Note: Descriptions are shown in the official language in which they were submitted.
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
SEGMENTATION AND STRATIFICATION OF COMPOSITE PORTFOLIOS
OF INVESTMENT SECURITIES
FIELD OF THE INVENTION
[001] The present invention relates generally to computerized techniques using
a logical
data model for constructing a stratified or segmented composite portfolio of
investment
securities.
BACKGROUND OF THE INVENTION
[002] The management of investment portfolios has been the subject of
substantial theory
and research. Portfolio theory considers how wealth should be invested and how
to
maximize a portfolio's expected return for a given amount of portfolio
liquidity-adjusted risk,
or, equivalently, minimize liquidity-adjusted risk for a given level of
expected return, by
carefully choosing the proportions of various assets. While a certain rate of
return may be
expected, the valuation of individual holdings in the portfolio can depart
upward or
downward from that expected rate of return. This upward and downward variation
from the
expected value is known as variance, or volatility. Over time, securities, in
theory, should
have an efficient frontier for expected volatility and return. According to
theory, securities
with a higher expected risk will have a higher expected return.
[003] Financial indices are often used to benchmark the performance of a
financial
instrument. The S&P 500 Index is an example of one such benchmark for stock-
oriented
funds and the Barclays Aggregate Bond Index is an example of a benchmark for
bond funds.
The S&P 500 is the largest equity benchmark in the world. Trillions of
dollars are either
invested in this benchmark or in funds benchmarked to it. Since yearend 1999,
U.S. broad
market indices such as the S&P 500 have experienced long periods of
underperformance.
For example, an investor in the S&P 500 at yearend 1999 was down
approximately 20% 10
years later in nominal terms at yearend 2009, depending on fees and treatment
of dividends.
It was not until late 2012 that the S&P 500 had a positive nominal return for
these yearend
1999 investors, including many large pension funds and endowments. As of
October 2014,
the S&P 500 had a negative real return since yearend 1999. During this same
period,
broad-based funds holding U.S. government or corporate debt have had positive
real returns
with corporate debt earning more than government debt during this period. This
premium
1
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
was due to the extra risk of a corporate bond versus a U.S. government bond of
comparable
duration. These markets had their annual fluctuations, but have been fairly
stable; over a
reasonable period of time, these securities had both positive returns and
differences that
would be expected based on risk. Neither of these statements can be made for
equity indices
such as the S&P 500 that lost value on an absolute basis and underperformed
materially
over a long period of time with respect to the less risky indices holding
investment-grade
corporate or government debt.
[004] Given a cap-weighted methodology, a change in the market value of a
relatively large
company has a disproportionate effect on an equity index, while a change in
the debt
outstanding of a relatively highly indebted issuer has a disproportionate
effect on a fixed
income index. Funds that track these indices also experience the corresponding
fluctuations
in value as the instruments representing the relatively larger companies
fluctuate in value.
[005] The S&P 500t, like most broad market indices, is capitalization-
weighted. This
means that the weight of an individual company in the index is proportional to
its market
capitalization relative to the other constituents. There are no controls in
the S&P 500 to
ensure that a single security or groups of securities that share a common risk
do not become
overweighted to represent too large a proportion of the portfolio. That is,
the types of
controls used in scientific fields and engineered processes where population
controls are used
to limit the influence that one part of a population can have on a total
population being
measured are not used in the broad market indices. Such controls limit both
positive and
negative influences. In population studies, controls are used to produce a
normative model of
an underlying population. Because there are no controls in the benchmarks
currently used to
invest in equity securities, there is no assurance that historical returns
from yearend 1999 to
the present are representative of equity securities in general. The strategy
of capitalization-
weighting without controls has produced below-average returns for long periods
of time.
[006] The results of the major U.S. broad-market equity indices since 1999
appear to be
inconsistent with the main theories of the pricing of investment securities
and the theory of
efficient markets. Much of the work on efficient markets and asset pricing
followed the
pioneering work of Markowitz and Sharpe with later notable additions by others
such as
Fama and French. Their theories suggest that individual securities are priced
at a level that is
expected to produce a risk-adjusted return relative to other investment
securities and that, by
following certain rules, a portfolio of securities has a higher probability
than an individual
2
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
security of achieving this risk-adjusted rate of return in any given period or
over several
periods.
[007] The principles that Markowitz and others proposed have been used to
assist investors
and managers in the selection of the most efficient portfolio design by
analyzing various
possible portfolios of a given set of securities. By delineating a portfolio
construction process
that entails choosing securities whose risk-return profiles diverge
significantly, the models
show investors how to reduce their risk. The foundational model in this area
is known as the
mean-variance model because it is based on expected returns (mean) and the
distribution
from expected returns (variance) of the various portfolios. When developing
the original
mean-variance model, Markowitz made the assumption that a portfolio that
maximizes return
for a given risk or minimizes risk for a given return is an efficient
portfolio. Thus, portfolios
are selected using the following rules: (a) from the portfolios that have the
same expected
return, the investor will prefer the portfolio with lower risk, and (b) from
the portfolios that
have the same risk level, an investor will prefer the portfolio with higher
expected rate of
return.
[008] To facilitate portfolio construction, Markowitz used the expected
covariance or
correlation among securities as an additional input that would enable
investors to maximize
their risk-adjusted return at the portfolio level. Although an individual
security may
underperform for a long period of time, the rules developed for efficient
portfolio
construction were designed to reduce, through diversification, this
probability of
underperformance with respect to the portfolio of securities. According to
these foundational
theories, investors could expect to be compensated only for systematic, or
broad-market,
risks, with a premium commensurate with the risks of a given asset class, and
should be able
to diversify away their exposure to non-systematic risks at the efficient
frontier, consisting of
the hypothesized market portfolio.
[009] One explanation for the inconsistency between modern portfolios and the
theoretical
portfolios on which the efficient market hypothesis was developed is that
modern portfolios
operate at a much greater scale and level of complexity than the theoretical
examples. The
early theoretical models based on the efficient market hypothesis and capital
asset pricing
model tend to use individual securities and describe diversification within
portfolios
consisting of numbers of securities that are in the single digits and low
double digits. Many
of the foundational papers were written before the mutual fund boom of the
1980s and 1990s
3
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
following the creation of individual retirement accounts (IRAs) by the
Employee Retirement
Income Security Act (ERISA) of 1974, as well as the introduction of the first
index fund in
1976. For example, the Markowitz paper on portfolio selection published in the
Journal of
Finance was written in 1952. According to the first shareowner census
undertaken by the
New York Stock Exchange (NYSE) in 1952, only 6.5 million Americans owned
common
stock at the time (about 4.2% of the U.S. population), and each held an
average of four
stocks. Sharpe's paper, "A Simplified Model for Portfolio Analysis," was
written in 1963
and his book, "Portfolio Theory and Capital Markets," was written in 1970,
long before the
mutual fund boom created by ERISA, the advent of globalization and modem
technology, the
development of exchange-traded products enabling retail investors en masse to
hold
thousands of securities at once, or the widespread recognition by
institutional investors of the
unique problems associated with managing such large funds.
[010] Modem portfolios manage trillions of dollars in the aggregate. The total
investment
into US mutual funds was $13 trillion dollars in 2012. In order to reduce
exposure to non-
systematic risks while avoiding relatively illiquid positions, the portfolios
require thousands
of securities in diverse risk groups. At this scale, lacking applicable
financial theory to guide
selections and weights, as portfolio theory was developed for portfolios of a
much smaller
scale, building efficient portfolios has been challenging. The absolute scale
of investment
today by very large institutions has grown exponentially since the mutual fund
boom of 80s
and 90s, discussed above. In addition, the underlying population of securities
has grown in
heterogeneity and complexity. This diversity and interconnectedness is
increasing every
year. The need to control for the non-systematic risks embedded in this
portfolio of
companies also increases every year.
[011] There is a need for a framework that enables the systematic comparison
and
contextualization of all types of securities in today's complex heterogeneous
global market.
Specifically, there is a great need for a framework that enables systematic
comparison and
contextualization of all types of equities in today's complex heterogeneous
global market. A
systems approach to organizing economic and financial information would
accomplish this
by enabling us to interrelate the vast data related to these activities and
analyze economic and
financial interdependencies.
[012] In addition, there is a need for a new normative methodology for
constructing
portfolios of investment securities, one that addresses the complexities of
today's companies
4
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
and the increasing size and diversity of today's funds by applying the
approach and
foundational principles of Markowitz and Sharpe to the complexities of today's
large-scale
funds.
[013] Some efforts for portfolio construction attempt to address the complex
heterogeneous
global market by relying on existing systems for classifying companies.
Current systems of
classification, such as Global Industry Classification Standard (GICS), are
not well-suited for
building new models of potential efficient portfolios of these large-scale
modern investment
vehicles that draw upon complex and globally interrelated universes of
equities. The NAICS
or GICS relate companies by their positions in a fixed hierarchy. There are
two significant
limitations of the fixed NAICS and GICS hierarchies: 1) any items without a
common parent
are unrelated and cannot be compared using terms in the hierarchy; 2) any
items sharing a
parent can only be compared along the terms that GICS or NAICS uses to label
that group
(insofar as the names of the groups indicate the term that separates them,
e.g., "consumer"
versus "commercial" may relate to the customer base).
[014] These systems, similar to the foundational papers in finance, were
created before the
advent of large digital databases; they are modeled after the frameworks of
the time such as
the Dewey Decimal System and Standard Industry Classification System. Those
systems rely
on a fixed hierarchy in which each entity has a single parent; that parent has
a single parent,
and so on. Each parent has descriptions, but not concepts of specific
attributes that would
enable an entity under one parent to be related to an entity under another
parent.
[015] Without the ability in the data structure to relate an entity under one
parent to an
entity under another parent, it is hard to understand the multivariate risks
to which companies
are exposed and, thus, to see how many securities in a large portfolio or
index may share a
similar or related risk. The shortcomings of current classification systems
are becoming
increasingly apparent given the complexities of today's companies and the
increasing size
and diversity of today's funds. Although many of the biggest risks in a
capitalization-
weighted strategy result from the lack of controls for single risk exposures,
bubbles, or
massive non-systematic price corrections, there are currently limited tools to
systematically
address these problems. Thus, there is a need for a multivariate attribute-
driven
categorization system enabled by current data processing capable of providing
these tools as
well as the ability to build multiple different portfolios to assess the
efficiency of each and
test for a normative case.
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[016] Benchmarks
[017] In addition to the systems used to organize securities and the
information about them,
modern portfolio construction is challenged by another step of the process
which has been
slow to evolve: the benchmarks against which to compare their performance. In
other areas of
economics and finance, the role of benchmarks has been well established.
Central banks
routinely use inflation targets to guide policy, which has proved instrumental
in increasing
the predictability of price changes. This has enabled consumers, merchants,
and investors to
consume, save, and invest with a high degree of confidence in near to medium-
term price
changes. National economic ministries routinely project their future
annualized GDP growth
and seek to achieve it, which multilateral institutions, banks, and investors
rely on as an index
of a country's economic health.
[018] In corporate finance, publicly traded companies regularly issue earnings
guidance
and have quarterly earnings targets, which it is the CFO's principal role to
achieve.
Companies are benchmarked against their earnings targets and held accountable
for them by
boards and financial analysts, and even minor shortfalls in earnings
frequently lead to
precipitous drops in stock price. CFOs are also expected to deliver on target
returns on
equity, which, since it is junior to debt in the capital structure, has a
higher cost of capital for
a given company and should have higher returns than the debt issued by a
company. In each
case, modern technology has enabled decision makers to more accurately
forecast future
economic and financial outcomes, control for risk, and achieve their
benchmarks with a high
degree of predictability.
[019] At the portfolio level, however, there is no comparable accountability
for equity
benchmarks. Since equity investments are riskier than debt investments at the
portfolio level
all equity indices should strive to earn a consistent premium to corporate
long-term bonds.
Just as all companies will expect a higher cost of equity than debt financing,
all equity
investors' indices, like the companies they invest in, should anticipate a
higher return when
they invest in a company's equity rather than its debt issuances. Because of
the statistical
properties of large sets of securities, investors should expect to see this
risk premium even
more consistently in an index portfolio. This risk premium should be realized
at the portfolio
level; equity index investors should strive to beat corporate long-term bond
returns for their
constituent group on a consistent basis.
6
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[020] The capital asset pricing model uses the term alpha to describe
outperformance of a
benchmark; from a company's perspective, generating alpha entails beating its
return
projections. For any given company, an equity premium is commensurate with
achieving
earnings estimates and outperforming borrowing rates. The same principle
should hold at the
portfolio and index level; investors in portfolios of equities should expect
returns that are
higher than the average borrowing rate for the bonds of a given constituent
group. If an index
or portfolio does not achieve the performance target predicted by theory, a
new methodology
is required that will realize that target more consistently and predictably.
[021] The S&P 500 is widely accepted as an equity benchmark even while it
continues to
lack risk controls and exhibit higher volatility than predicted by theoretical
models. It fails to
achieve the rates of return predicted for it by the foundational finance
theories and asset
pricing models. Nevertheless, the methodology of the S&P 500 has not changed
significantly
since its inception, and it has failed to capitalize on the tools of modern
technology and data
analytics to control for risk and achieve more predictable, reliable rates of
return. Thus, there
is a need for a reconsideration of how to construct equity benchmarks and the
standards for
them.
[022] Conglomerates
[023] Corporations have sought to achieve diversification at the company level
through the
conglomerate form, which involves acquiring and managing multiple
independently operated
and often functionally unrelated businesses through a parent company. Owners
of
conglomerates sought to reduce the volatility in earnings associated with
business cycles in
various industries by organizing relatively uncorrelated income streams under
the same
corporate structure; some also sought to achieve cost savings through
synergies in
procurement, branding, marketing, and sales, to avoid antitrust restrictions
on expansion and
consolidation in a particular industry by aggregating interests across
multiple sectors.
[024] Although conglomerates have enjoyed substantial popularity in certain
wealthy
countries following long periods of high economic growth--the U.S. in the
1960s, Japan in
the 1980s, and more recently, South Korea--they have largely fallen out of
favor in high-
income markets. The extra layers of bureaucracy and lack of sufficient
industry expertise at
the holding or parent company level frequently have made conglomerates too
complex to
manage effectively.
7
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[025] More recently, private equity firms have sought to achieve similar
objectives to those
of conglomerate managers by acquiring and managing mature businesses,
frequently in a
wide variety of industries. The significant fees charged by such firms,
typically comprising
2% of assets managed and 20% of returns over a benchmark in addition to deal-
specific fees,
have impeded their ability, as a group, to generate high returns to investors,
while other firms
have foundered due to similar challenges that confronted conglomerates, failed
to capitalize
on potential marketing, sales, and operational synergies, or incurred
excessive leverage that
contributed to large losses during economic downturns.
[026] While some private equity firms consistently have shown very strong
performance,
most of them are limited partnerships inaccessible to the general public due
to regulatory
restrictions, and the information regarding their operations, strategy, and
investments is
largely opaque and frequently unavailable. The lack of transparency and
liquidity in these
funds, as well as the challenges involved in managing businesses across
disparate sectors,
have impeded the capacity of these firms to scale. At present, the largest
traditional
investment firm itself manages more capital than the entire global private
equity industry
combined.
[027] Volatility
[028] Volatility in pricing refers to fluctuations in price. Volatility is a
significant factor in
portfolio performance and these price fluctuations may create a drag on
portfolio growth. For
example, daily volatility has been shown to hurt the return of leveraged
exchange-traded
funds. Random movements in investment securities without controls at the
portfolio level,
especially large downward movements caused by unpredictable events or the
popping of non-
systematic bubbles, reduce risk and liquidity-adjusted returns. In these
cases, there is little to
no expectation that portfolios and their constituent investment securities
will rebound to pre-
existing levels. In both of these cases, the securities being impacted are
being re-priced
because of new information or a sudden market recognition that they were
overpriced.
[029] In an effort to reduce the effects of volatility on a portfolio, various
weighting
schemes have been proposed in the investment industry. For example, one method
described
in U.S. Patent No. 8,306,892 operates by calculating weights based on market
capitalization,
gross-domestic product, and geographic region. In another example, described
in U.S. Patent
No. 8,131,620, weights in a portfolio of securities are based on market
capitalization and
dividend yield. Numerous other portfolio weighting schemes exist. However,
none of these
8
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
weighting schemes fully address the shortcomings of prior art portfolio
theory, as discussed
above. Some examples, such as that described in U.S. Patent No. 8,005,740, use
accounting-
based metrics for weighting securities universes.
[030] In prior art portfolio construction, random groups of securities are
likely to have
periods of significant valuation swings, both up and down, from one time
period to another.
These massive swings in value in random groups of securities may not be caused
by variables
such as accounting attributes or their designation as "growth" or "value"
stocks. The
valuation swings could be caused by, for instance, companies being long a
specific
commodity when the commodity suddenly loses its value; over-exuberance in the
demand
prospects for a company's or industry's product that does not meet demand;
long fixed-cost
contracts when the actual costs available to their competitors changes; over-
weighting of a
certain asset in the product mix when that asset loses its value; or other
idiosyncratic reasons.
[031] There are many reasons for apparently random bubbles. In some cases,
they are
systematic or broad-market bubbles; in others, they are largely limited to a
constituent group
(such as an asset class or industry). There are certain events that appeared
to be systematic
because they impacted index and portfolio returns so severely, such as the
Internet bubble of
the late 1990s, but are non-systematic. In either case, the impact on an
investor's returns
when the bubbles collapse can be extremely negative as a result of portfolio
biases and
overexposure to constituents that are especially impacted by the collapse of
the bubble.
[032] The random walk hypothesis in financial theory represents the inability
to address the
apparent randomness of volatility and returns in equity-based investment
securities. The
hypothesis implies that in an efficient market, a large random selection of
equity-based
investment securities will perform as well as an actively-managed selection of
equity-based
securities, before adjusting for taxes and fees. The random walk hypothesis is
the underlying
reason for the proliferation of index funds and the broad support for passive
index funds by
the academic community. The hypothesis, taken to its logical extreme, suggests
that a
blindfolded monkey throwing darts at the stock listings could select a
portfolio that would do
just as well as one selected by the experts.
[033] Many different weighting strategies have been proposed to deal with this
problem of
random volatility in equity-based investment securities. The recent
underperformance of
these passive capitalization and even-weighted indices to debt indices that
track comparable
9
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
universes of companies has highlighted that these passive indices continually
affirm the same
randomness hypothesis.
[034] A major problem in the risk management of large portfolios of securities
is the
inability in existing systems to control for the occurrence of these types of
events without a
framework to define homogeneous subpopulations. If a portfolio inadvertently
over-weights
in a security or groups of securities that have a common bubble or bankruptcy
risk, the
returns can be materially impacted by a relatively small number of securities
in the portfolio.
Non-systematic bubbles and bankruptcies are associated with non-systematic
factors of the
industries, companies, or assets associated with specific investment
securities. In several
cases, over-weighting in specific non-systematic variables has caused
significant negative
impacts on a portfolio. This was clearly the case of the Internet bubble. In
calendar year
2000, the capitalization-weighted S&P 500 was down 9.09%. In that year, there
were 16
stocks that were down 49.8%, while the rest of the market was up 4.28%.
Unfortunately for
investors in funds tracking that index, these 16 companies, which were all in
the business of
moving, storing, or processing information, comprised 24.8% of the total
portfolio. The
underperformance of these select securities had a massively disproportionate
effect on the
index, and the trillions of dollars in funds benchmarked to it, because of the
lack of controls
on the underlying index.
[035] Prior efforts to improve portfolio returns generally appear to have at
least three
problems: 1) a sub-optimal number of groups; 2) insufficient ability to
control for covariance
within groups or correlation among groups to ensure that each group operates
in a predictable
group-specific way; and 3) no way of defining a group in a systematic way that
is applicable
across an entire economy and permits all groups to be related to one another.
Existing large-
scale heterogeneous indices and portfolios of securities lack controls on
their constituent
groups and neither capitalization-weighting nor even weighting are capable of
reducing the
impact of group-specific risks at the portfolio level in a population of
securities.
[036] Covariance and Correlation
[037] While finance theorists have made significant breakthroughs in
forecasting the return
and variance for individual securities, there has been little advancement in
finding reliable
indicators of the pairwise correlations or covariances between securities, a
required input to
the Markowitz model. In 1973, financial economists Edwin Elton and Martin
Gruber
addressed why quantitative solutions are unlikely to be practicable at scale,
and noted that to
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
obtain efficient portfolios from among 200 stocks, 19,990 correlation
coefficients would have
to be produced.
[038] There are also institutional impediments to finding generally applicable
and
sufficiently explanatory indicators, as there is highly unlikely to exist any
individual at a
financial institution sufficiently familiar with the mathematical analysis of
each constituent of
a substantial equity universe to be able to approximate a quantitative
solution. Elton and
Gruber concluded that there is no non-overlapping organizational structure
that would permit
security analysts in a financial institution to produce estimates of
correlation coefficients
between all relevant pairs of stocks, since each analyst follows a subset of
the stocks in which
the institution has an interest.
[039] In an effort to address the lack of reliable indicators of the
correlation in how
securities perform, traditional models such as the capital asset pricing model
(CAPM) assume
that all residual pairwise correlations are zero. That is, it is assumed that
each security has no
relationship to any other security in excess of co-movement with the market as
a whole. This
assumption lacks realism: a simple likelihood ratio test for zero correlations
rejects the null
hypothesis of zero residual pairwise correlations at the 0.000001 significance
level.
[040] Elton and Gruber illustrate that the CAPM can be improved upon simply by
assuming
a single nonzero pairwise correlation to be assigned across an entire
portfolio, but
acknowledge the severe limitations of this approach. The challenges referenced
above, and
the lack of a well-developed, field-specific framework to address the
covariance issue at
scale, have left the problem unsolved. The increasing scale, complexity, and
heterogeneity of
modern portfolios have made this challenge more acute.
[041] Purely quantitative measures of correlation have proven least accurate
and least
predictive precisely when they are most needed: during bubbles, crashes, and
other periods of
high market volatility, when these measures have deviated far from their
historical norms.
Investors who have sought to diversify principally based on quantitative
historical
covariances have sustained extraordinary losses during recent periods of
market volatility.
[042] Factor Models
[043] Asset pricing models such as the CAPM frequently have failed to
accurately describe
or predict performance characteristics of securities, groups of securities, or
portfolios. These
models isolate a very small number of factors believed to be driving security
price returns
and are predicated on the assumption that they can be determined purely
quantitatively.
11
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[044] The CAPM relies on the risk free rate, the market return, and the
idiosyncratic risk of
the security; in other words, it is predicated on the assumptions (among
others) that there is
one factor F common to all securities in the market, there exist a set of
factors f1,2...nwhich
map precisely, in a one-to-one correspondence, to the set of securities
s1,2...,,, that these
factors and their weights are essentially stable over time, and that the
relationship among
these factors and their weights is entirely unknown.
[045] The Fama-French three-factor model adds size and book to market value to
the
aforementioned factors, while their posited five-factor model, which, as of
November 2013,
also adds profitability and asset growth, does not yet appear to improve on
their previous
model. (Eugene Fama and Kenneth French, "A Five-Factor Asset Pricing Model,"
working
paper, September 2014.) Carhart's posited four-factor model adds momentum to
the three-
factor model. (Carhart, M. M., "On Persistence in Mutual Fund Performance,"
The Journal of
Finance 52: 57-82 (1997).) Tobias Adrian, Emanuel Moench, and Hyun Song-Shin
point to
the systemic impact of aggregate broker-dealer capital structure and asset
growth in non-
banking financial institutions on equity and bond prices. (Tobias Adrian,
Emanuel Moench,
and Hyun Song Shin, "Financial Intermediation, Asset Prices, and Macroeconomic
Dynamics," Federal Reserve Bank of New York, 2010.) Andrew Lo and Amir
Khandani add
common factors such as general market volatility and commodity prices, and
emphasize
liquidity as an additional factor at the security level which was unduly
neglected in studies of
large and mid-cap stocks in developed markets during periods of little
turbulence, when
liquidity factors are less relevant. (Andrew Lo and Amir Khandani,
"Illiquidity Premia in
Asset Returns," draft paper, June 2009.)
[046] Methodologies focusing first on quantitative analysis that have failed
to identify any
factors or risks other than systematic and idiosyncratic, or the relationship
among the various
idiosyncratic factors or risks, and a lack of computing power when many of the
key
paradigms of finance were formulated, have led portfolio and index
construction to be
predicated on the assumption that all drivers of security price returns either
a) affect every
security in the entire market precisely the same way, or b) affect only one
security in the
entire market in any way at all. This untenable assumption has made effective
portfolio and
index construction extremely difficult.
12
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[047] Problems of Scale
[048] For multiple reasons, the problems described above are particularly
acute in large-
scale portfolios of securities. Various reasons why management at scale is
difficult are
provided below.
[049] (a) Charter limits on ownership: For many funds and fund managers, there
are limits
on the percentage of a company they can own. For example, for any fund that
seeks to
acquire a 5% holdings of U.S. public equities, there are required 13-D filings
and more
extensive regulatory oversight. Many funds will not or cannot cross that
threshold.
[050] (b) Liquidity limits on ownership: The more a fund owns of an individual
security,
particularly for large holdings, the harder it generally is to sell. The
effect is frequently trivial
for small dollar value holdings in liquid securities, but may be significant
for larger holdings
or relatively illiquid securities.
[051] (c) Large funds need a large number of securities to fill out a
portfolio: Due to the
factors identified above as well as other practical issues, a large fund needs
a large number of
companies to invest in due to liquidity and ownership issues. Across an
economy, there are
many linkages among companies, and the larger the number of companies under
evaluation,
the more difficult it is to track and oversee the linkages and risks that come
from them.
[052] (d) Large funds may face a limited selection of securities: Due to the
factors identified
above as well as many more practical issues, large funds often need to invest
disproportionately in large companies or other funds. The available companies
in this group
vary over time. In addition, these securities have variable weights and
aggregate differently
depending on what companies exist in which category at any given point in
time.
[053] (e) Geographic variation: In addition to changes over time, this
industry, sector, or
company selection varies by geography; in large portfolios, indices, or funds
comprised of
securities, determining the geographic exposure of assets, operations, and
products, as non-
limiting examples, is impracticable using prior art methods. Sector
differentiation may be a
greater cause of price movements between geographies than the underlying
currency that
drives the products. For example, portfolios of US securities are often more
heavily weighted
in technology stocks than portfolios of European or Latin American securities.
Europe and
Latin America are relatively heavy in manufacturing and financials.
13
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[054] If a fund, index, or portfolio manager's goal is currency
differentiation, it is important
to control for these sector variations. Not only understanding the different
potential risk
groups that exist at any given point in time and in any specific geography or
category, but
also being able to control for these risks is difficult using currently or
previously known
techniques.
[055] (e) Attribute and overconcentration risk are multi-dimensional: Single
and multiple
attributes are helpful in distinguishing risks in individual companies, but
attributes that are
clear on an individual level are lost in larger classification systems. These
varied, yet
critical, attributes impacting security price returns are often aggregated
into one technology
metacategory in large-scale funds. The existing categories in current systems
tend to be
standardized on a global basis and do not permit differentiation among these
attributes that
aggregate to characterize each metacategory. The inability to represent linked
multi-attribute
risks is a significant limitation for existing large-scale investment
portfolios.
[056] If portfolios, and large-scale portfolios in particular, are not better
controlled, and the
linkages between companies are not well understood, non-systematic events can
appear to
have systematic impact. Examples of non-systematic events are provided below.
Known and
existing classification systems do not address the underlying statistical
causes for the
systematic impact of the volatility of the constituents of large-scale
portfolios of securities.
With improved controls, however, the impact of non-systematic events could be
limited.
BRIEF SUMMARY OF THE INVENTION
[057] Without both a reliable and validated classification system using
functional attributes
as well as a computerized system that uses a stratified or segmented (or
blocked) composite
structure, prior art systems are unable to control for the different
attributes associated with the
securities. A stratified or segmented composite portfolio can be formed by
selecting a group
of investment securities, segmenting the securities into sub-groups according
to attributes that
correlate to one or more identified investment security risks, and assigning
portfolio weights
to one or more sub-groups based on their stratified or segmented positions.
The attributes can
be selected from a universe of possible values. Further positive and negative
biases can be
applied at any arbitrary point, stratum, or segment, including to individual
investment
securities, groups of arbitrarily selected investment securities, or arbitrary
positions in the
architecture.
14
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[058] The specific functional attributes associated with the investment
securities can be
used to segment, stratify, and weight the holdings of investment securities in
a portfolio by
assigning specific weights to the risk groups in which the underlying
securities are held in
order to meet the engineered risk objectives of the overall portfolio. As a
non-limiting
example, one of the goals in segmenting or stratifying risk groups may be to
reduce the
impact of attribute-specific volatility drag on the portfolio as a whole. As
non-limiting
examples, the systems and methods described herein can be used in investment
management
by controlling for specific types of random events that impact the overall
randomness of risk,
return, skewness, and kurtosis in large portfolios or groups of investment
securities.
[059] Multi-attribute risk composites can provide a tool to manage risk by
reducing or
minimizing the potential risk resulting from these attributes and/or
increasing or maximizing
the potential return from these type of risks by engineering the composite to
take advantage
of an event a manager expects to happen.
[060] In some embodiments, a stratified composite portfolio can be created by
tagging
securities with risk attributes based on functional attributes and applying a
weighting scheme
that limits the exposure to individual attributes. The result of this process
is a weighted
portfolio that stratifies or segments risk exposure across a number of risk
attribute categories,
and disperses the risk in the individual groups and sub-groups according to
attribute
categories within groups, to achieve a desired risk profile that can be
represented by a target
score.
[061] In one aspect of the disclosure, there is provided a computer-
implemented method for
storing a representation in a database of an index or portfolio of investment
securities, the
method comprising electronically storing one or more data entities in a
database system, each
of the data entities representing the identity of an investment security, the
investment security
associated with a corresponding economic entity; electronically tagging each
data entity with
one or more functional attributes of the corresponding economic entities;
wherein the
functional attributes characterize the roles of each of the economic entities
in one or more
processes converting inputs to outputs; selecting multiple investment
securities represented
by the data entities for inclusion in an index or portfolio of investment
securities; defining at
least a first group and a second group of investment securities based on the
electronic tags or
the functional attributes associated with the corresponding economic entities;
segmenting the
selected investment securities into the two or more groups based on the
electronic tags or the
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
functional attributes; wherein the investment securities in the first
segmented group share a
first common or proximate functional attribute, and the investment securities
in the second
segmented group share a second common or proximate functional attribute;
electronically
accessing the database representation of the segmented groups; electronically
iterating
through the accessed representations to compute a negative or positive weight
for one or
more of the investment securities based on the one or more segmented groups
into which the
investment securities are segmented; and assigning the negative or positive
weight to the one
or more of the investment securities; and electronically storing the assigned
weight in the
database system.
[062] Further embodiments include selecting one of the segmented groups of
investment
securities which share a first common or proximate functional attribute;
segmenting the
selected group of investment securities into two or more sub-groups, wherein
the sub-groups
are subsets of the segmented groups; weighting the two or more segmented sub-
groups;
wherein the investment securities in a first sub-group share a third common or
proximate
functional attribute and the investment securities in a second sub-group share
a fourth
common or proximate functional attribute.
[063] In further embodiments, the joint intersection of each set of groups is
the empty set;
and the joint intersection of each set of sub-groups is the empty set. In
further embodiments,
one or more groups, sub-groups, or investment securities are weighted based on
syntactic or
functional tags, or syntactic or functional attributes; and one or more
groups, sub-groups, or
investment securities are weighted based on non-syntactic, non-functional
tags, or non-
syntactic or non-functional attributes.
[064] Further embodiments include assigning a target weight to a group, sub-
group, or
investment security; and periodically rebalancing the group, sub-group, or
investment
security to the target weight.
[065] In further embodiments, one or more portfolios, indices, groups, sub-
groups, or
securities, or the data entities representing them, are represented in
graphical, sequential,
clustered, or networked form.
[066] Further embodiments comprise electronically using predictive analytics
based on
functional attributes to forecast the performance, volatility, liquidity,
variance, expected
return, alpha, Jensen's alpha, beta, variance, covariance, semivariance,
semideviation,
16
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
correlation, autocorrelation, Sharpe ratio, Sortino ratio of one or more
portfolios, groups, sub-
groups, or investment securities, or the excess or residual of any of these
metrics.
[067] In further embodiments, one or more weights are assigned to an
investment security
based on a functional attribute of a corresponding economic entity, or
electronic tag
representing such an attribute.
[068] Further embodiments comprise transmitting, sending, or relaying
information
regarding one or more data entities and one or more weights to an exchange,
index provider,
index calculator, brokerage, asset manager, investment advisor, investment
manager,
specialist, broker-dealer, authorized participant, trader, financial
professional, investment
professional, investor, general partner, limited partner, private equity
investor, venture capital
investor, hedge fund investor, conglomerate manager, executive, pension fund
advisor,
endowment manager, fund manager, or securities trading platform.
[069] Further embodiments comprise using one or more weights to construct an
index, buy,
sell, issue or transmit an order, or execute trades in an investment security,
group, or
portfolio.
[070] In further embodiments, the functional attributes are associated with
risk exposures,
and wherein at least two groups of investment securities are associated with
different
functional attributes and different risk exposures.
[071] Further embodiments comprise associating two or more numerical values
with two or
more groups, tags, attributes, risk exposures or relationships, wherein the
numerical values
relate to economic, financial, or capital markets-based data; associating a
statistical property,
selected from among mean, variance, standard deviation, skew, kurtosis,
correlation,
semivariance, and semideviation, with those groups, tags, attributes, risk
exposures, or
relationships based on the numerical values; calculating two or more
statistical values
associated with the statistical property; determining the statistical
significance of the
calculated statistical values of each group, tag, attribute, risk exposure, or
relationship;
validating that the statistical values are significant at a predetermined
level; and if the values
are not significant, reassigning groups, tags, attributes, risk exposures, or
relationships.
[072] In further embodiments, the number of securities in each group is chosen
such that a
statistical power of the statistical test exceeds a predetermined level.
17
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[073] In further embodiments, the investment securities or groups are selected
from among
equity, debt, derivatives, currencies, commodities, funds, notes, alternative
investments,
exchange-traded products, real assets, and structured products.
[074] Further embodiments comprise selecting a financial or economic metric to
measure
with respect to one or more of the groups, indices, or portfolios, wherein:
the distribution of
expected or realized values of the metric for the index, portfolio, or group
is relatively more
normal than the distribution of expected or realized values of the metric for
an alternative
index, portfolio, or group; or the value of the metric is more stable or
predictable for the
index or portfolio than it is for the group, as measured by a mathematical
test of stability or
predictability; or the value of the metric is more stable or predictable for
the group than it is
for an investment security, as measured by the mathematical test of stability
or predictability.
In some further embodiments, the normality of the distribution is assessed
using Cramer¨von
Mises criterion, Kolmogorov-Smirnov test, Shapiro-Wilk test, Anderson-Darling
test, Jarque-
Bera test, Siegel-Tukey test, Kuiper test, p-value test, a Q-Q plot, a test of
skewness, or a test
of kurtosis. As non-limiting examples, stability may be assessed through a
test of variance or
a test of heteroscedasticity.
[075] Further embodiment comprise electronically storing one or more data
entities, each of
the data entities representing the identity of a segmented group, the
segmented group
comprising one or more investment securities and associated with one or more
corresponding
economic entities; and electronically tagging each group with one or more
functional
attributes of the corresponding economic entities.
[076] Further embodiments comprise identifying an index or benchmark to track;
selecting,
grouping, or weighting the investment securities so as to track substantially
or replicate the
performance of the identified index or benchmark.
[077] In further embodiments, the portfolio or index comprises a synthetic
conglomerate.
[078] In further embodiments, one or more weights are assigned based on
semantic,
syntactic, morphological, morphosyntactic, anatomical, physiological,
functional, graphical,
or value chain proximity.
[079] Further embodiments comprise electronically storing a computerized
representation of
an economic systems syntax, wherein the economic systems syntax can be applied
by a
computer processor to establish the validity of expressions of elements of the
system based
on one or more functional properties of the economic entities.
18
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[080] Further embodiments comprise recommending a portfolio, group, or
investment
security to a user based on functional attributes electronically identified by
the system or
user.
[081] Further embodiments comprise arranging the selected data entities into a
stratified
structure including at least two parent groups and at least two sub-groups of
each parent
group such that: one or more parent groups are defined by one or more
functional attributes
such that data entities of those parent groups have in common the attributes
that define those
parent groups and wherein at least two parent groups are associated with
different risks; the
sub-groups inherit one or more functional attributes and corresponding risks
from the parent
groups; and the sub-groups are defined by one or more divergent functional
attributes such
that one or more sub-groups are associated with different risks from the
parent groups and
from other sub-groups.
[082] Further embodiments comprise calculating a measure of statistical
dependence
between each of two parent groups and between each of two sub-groups;
determining
whether the parent groups and sub-groups have relatively high intra-group
statistical
dependence; determining whether the parent groups and sub-groups have
relatively low inter-
group statistical dependence; and if the intra-group statistical dependence
does not exceed the
inter-group statistical dependence, reorganizing the groups or sub-groups.
[083] In further embodiments, one or more sub-groups are assigned weights
relative to one
another according to a weighting scheme such that the weight of one or more
parents equals
the sum of the products that result from multiplying a sub-group by its
assigned weight
according to the weighting scheme.
[084] In further embodiments, the realized returns of the portfolio exceed
those of a
commercially available index or benchmark, over the previous one, three, and
five years for a
given level of risk, or match those of the index or benchmark at a lower level
of risk; wherein
the securities in the portfolio are the same, or substantially the same, as
the securities in the
index or benchmark.
[085] In another aspect of the disclosure, there is provided a computer-
implemented method
for storing a database characterization of an index, portfolio, set,
aggregate, or composite of
elements of a functional system, or of a representation of those elements, the
method
comprising: electronically storing a set of data entities in a database
system, each of the data
entities corresponding to an element of a functional system; wherein the
functional system
19
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
comprises a group of elements ordered by their functional roles in converting
inputs to
outputs, or as the inputs, or as the outputs; electronically assigning each
data entity associated
with an element one or more functional attributes represented as an electronic
tag; wherein
the functional attributes characterize the roles of each of the elements in a
process of
converting inputs to outputs; selecting multiple elements, or a representation
of those
elements, characterized by data entities for inclusion in a portfolio, index,
set, aggregate, or
composite; segmenting the selected elements, or a representation of those
elements, into two
or more defined groups based on the electronic tags representing the
functional attributes
associated with the corresponding elements; wherein the first group shares a
first common
functional attribute, and the second group shares a second common functional
attribute;
electronically accessing the database representation of the segmented groups;
electronically
iterating through the accessed representations to compute a negative or
positive weight for
one or more of the elements, or a representation of those elements, based on
the one or more
segmented groups; and assigning the negative or positive weight to the one or
more of the
elements, or a representation of those elements; and electronically storing
the assigned weight
in the database system.
[086] In further embodiments, the functional system is economic; the elements
comprise
one or more inputs, outputs, resources, activities, functions, businesses,
enterprises, jobs,
companies, projects, products, assets, shareholder's equity, liabilities,
commodities,
currencies, imports, exports, communities, or economic interests in, or
collections of, any of
the foregoing, in the economic system; investment securities represent the
elements of the
economic system; wherein one or more investment securities, or one or more
groups, are
selected from among equity, debt, derivatives, currencies, commodities, funds,
notes,
alternative investments, exchange-traded products, real assets, and structured
products; and
one or more data entities identify one or more investment securities.
[087] In another aspect of the disclosure, there is provided a system for
executing a
command in a computing environment to construct a representation of an index
or portfolio
of investment securities in a database, the system comprising: a computerized
processor
configured for: electronically tagging one or more data entities with one or
more functional
attributes of the corresponding economic entities; wherein the functional
attributes
characterize the roles of each of the economic entities in one or more
processes converting
inputs to outputs; selecting multiple investment securities represented by the
data entities for
inclusion in an index or portfolio of investment securities; defining at least
a first group and a
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
second group of investment securities based on the electronic tags or the
functional attributes
associated with the corresponding economic entities; segmenting the selected
investment
securities into the two or more groups based on the electronic tags or the
functional attributes;
wherein the investment securities in the first segmented group share a first
common or
proximate functional attribute, and the investment securities in the second
segmented group
share a second common or proximate functional attribute; electronically
accessing the
database representation of the segmented groups; electronically iterating
through the accessed
representations to compute a negative or positive weight for one or more of
the investment
securities based on the one or more segmented groups into which the investment
securities
are segmented; and assigning the negative or positive weight to the one or
more of the
investment securities; and an electronic data store configured for:
electronically storing the
one or more data entities in a database system, each of the data entities
representing the
identity of an investment security, the investment security associated with a
corresponding
economic entity; electronically storing the assigned weight in the database
system.
[088] In further embodiments, the computerized processor is further configured
for:
selecting one of the segmented groups of investment securities which share a
first common or
proximate functional attribute; segmenting the selected group of investment
securities into
two or more sub-groups, wherein the sub-groups are subsets of the segmented
groups;
weighting the two or more segmented sub-groups; wherein the investment
securities in a first
sub-group share a third common or proximate functional attribute and the
investment
securities in a second sub-group share a fourth common or proximate functional
attribute.
[089] In further embodiments, the computerized processor is further configured
for:
selecting one of the segmented groups of elements, or representations of those
elements,
which share a first common or proximate functional attribute; segmenting the
selected group
of elements, or representations of those elements, into two or more subgroups,
wherein the
subgroups are subsets of the segmented groups; wherein the groups of elements,
or
representations of those elements, in a first subgroup share a third common or
proximate
functional attribute and the elements, or representations of those elements,
in a second
subgroup share a fourth common or proximate functional attribute.
BRIEF DESCRIPTION OF THE DRAWINGS
[090] FIG. 1 illustrates an example method for creating a stratified composite
portfolio and
weighting investment securities.
21
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[091] FIG. 2 illustrates an example method for creating a stratified composite
portfolio and
weighting investment securities.
[092] FIG. 3 illustrates an example stratification with three levels.
[093] FIG. 4 illustrates an example data set consistent with the example three-
level
stratification.
[094] FIG. 5 illustrates an example method for creating a stratified composite
portfolio and
weighting investment securities.
[095] FIG. 6 illustrates an example method for calculating weightings for a
stratified
composite portfolio.
[096] FIG. 7 illustrates an example method for creating a stratified composite
portfolio with
a target score.
[097] FIGs. 8A-8B illustrate an example architecture represented as statements
defining an
architecture and barcode.
[098] FIG. 9 illustrates example relationships between syntax elements
graphically.
[099] FIG. 10 illustrates an example database implementation for the system.
[100] FIG. 11 illustrates an example computerized system for stratified
composite portfolio
weighting.
DETAILED DESCRIPTION
[101] Definitions
[102] Investment Security: As used herein, an investment security is defined
as a financial
instrument that can represent, as non-limiting examples: an ownership position
in a
corporation (stock), a commodity, or a collection of assets; a securitized
creditor relationship
with an institution, such as a corporation, multilateral, or governmental body
secured directly
or indirectly by the assets of the issuer (bond); potential rights of
purchase, sale, or ownership
as represented by an option or other derivative instrument; a security
interest in a commodity
or real asset, including, as non-limiting examples, energy, timberland, and
precious metals; a
group of other securities pooled into a security, including, as non-limiting
examples, a fund,
exchange-traded fund, exchange-traded product, or structured product; or any
collection
thereof A security may be a fungible, negotiable, financial instrument that
represents a type
of financial value associated with an economic entity. The company or economic
entity that
22
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
issues the security is known as the issuer. The value of the security value
can be based on the
type of security, the type of relationship with the issuer, and the type of
assets and liabilities
that are directly or indirectly associated with the security.
[103] Economic entity: As used herein, an economic entity is involved in some
capacity,
whether active or passive, in the production, distribution, trade, or
consumption of real or
virtual goods or services. As non-limiting examples, an economic entity may be
a
corporation, company, enterprise, business, work group, department, laborer,
input, output,
resource, activity, function, project, product, assets, liability,
commodities, currencies,
imports, exports, community, job, worker, individual, governmental body,
intergovernmental
organization, multilateral organization, non-governmental organization, social
enterprise,
charity, non-profit, or any collection thereof As non-limiting examples, an
economic entity
may pursue financial, environmental, social, or governmental objectives, or
some
combination thereof
[104] Functional Attributes: The economic entities represented by the
investment security
can be associated with or have attributes. Functional attributes characterize
the roles of the
economic entities in processes converting inputs to outputs. The database
system can operate
on multiple types of attributes associated with an entity. As non-exclusive
examples, the
database system can operate on classes of attributes that are: (a) relative,
and/or (b)
functional, and/or (c) contextual; and/or (d) absolute. Relative attributes
may be, for
example, syntactic attributes, geographic, temporal, scoring systems,
designations as
high/low volume securities or as growth/value securities.
[105] In some cases, attributes can be defined to include attributes relating
to the entity
associated with the security and correspondingly exclude attributes of the
security itself For
those embodiments, the database system can be configured to define attributes
so as to
specifically exclude attributes relating to the type of investment security,
such as equity, debt,
or derivative, and characteristics of the investment security, such as
preference, maturity,
duration, or strike price. In those configurations, those excluded attributes
are not considered
to be functional attributes because the included attributes relate to the
economic entity with
which the investment securities are associated, not the security itself
[106] In some embodiments, functional attributes can be applied to inputs,
outputs, or
functions transforming inputs to outputs. In other embodiments, functional
attributes may
apply to activities, resources, systems, subsystems, composites, or elements.
As non-limiting
examples, types of functional attributes can be: syntactic attributes or
semantic attributes.
23
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
Examples of functional attributes may include, as non-limiting examples: (a)
attributes
related to investors (e.g. "institutional" v. "non-accredited"), (b)
attributes of assets belonging
to the company (e.g., "outsourced" vs. "in-house" for a manufacturing
company), (c)
attributes related to the product (e.g., "raw material" vs. "simple
component"), (d) attributes
related to customers (e.g., "business" v. "consumer" v. "government"), and (e)
attributes
related to suppliers (e.g. "wholesale" v. "retail"). The system can recognize
any combination
of different types of attributes. Some attributes may have qualities that are
both relative-to-
universe and functional.
[107] In some embodiments, functional attributes can be defined to exclude
accounting and
performance-based attributes. In some embodiments, the functional attributes
can be
qualities, features, properties, or inherent characteristics of the underlying
entity or assets
with which an investment security is associated. Functional attributes may
define
relationships throughout the value chain and structure of an economic entity,
including, as
non-limiting examples: (a) what a company does, such as manufacturing or
transportation;
(b) aspects of the company's product, such as specific utility provided by the
car, computer or
couch; (c) what the company's customer does, such as consumer sales or
business
intelligence; (d) what the customer's customer does; (e) the products and
materials a
company uses to provide its product; (0 the multivariate industries or
industry segments in
which a company may operate; (g) the structure of a company's business, such
as integrated,
non-integrated, forward integrated, backward integrated or networked; (h)
risks based on a
company's management, including its decisions and strategies; (i) risks based
on the internal
operations of a company.
[108] A major part of the linkages in an economic system are due to non-
systematic
functional attributes associated with, as non-limiting examples, a company's
suppliers,
products, industry, and operations, and geographic location. Without a
comprehensive
awareness of such shared attributes or linkages, it is very easy for
portfolios with a large
number of securities to become over-concentrated in non-systematic risk
categories.
[109] At any given point in time, any one of these attributes or an industry
event related to
these attributes may affect the risk associated with securities associated
with entities that have
these attributes. Understanding different potential risk groups and
controlling for them is
difficult without both a reliable and validated system of functional
attributes as well as a
stratified or segmented composite architecture to control for the different
attributes.
24
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[110] Functional attributes may be syntactically or semantically structured;
they can be
framed in natural or symbolic, relational language using the methods described
in U.S Patent
Application Ser. No. 14/216,936, the contents of which are hereby incorporated
by reference
herein. Any combination of multiple attributes can be formed as a compound
attribute.
Compound attributes can be defined as a new single attribute.
11111 Stratified Composite Unit: As used herein, a stratified composite unit
is defined as a
stratified set of investment securities comprising: 1) a parent group that is
defined by one or
more attributes where all members of the parent group have in common the
attributes used to
define the parent group; and 2) at least two sub-groups of the parent group,
which may be
considered to be children of the parent group and/or siblings of each other.
All members of a
sub-group have in common the attributes used to define the sub-group and its
parent group.
Any stratified composite unit and its constituent sub-units can include an
arbitrary number of
other sub-units that follow the rules of its parent unit or sub-unit. In some
cases, a stratified
composite unit may be comprised of only a parent group and two sub-units. In
other cases, a
stratified composite unit may be comprised of as many parts as the size and
diversity of the
original parent will support. With reference to Fig. 4, a stratified composite
unit can
comprise elements 1210, 1230, and 1235.
[112] Segmented Composite Unit: As used herein, a segmented composite unit is
defined as
a segment for securities comprising: 1) a segmented group defined by one or
more shared
attributes; 2) at least two sub-segments of the larger segmented group, each
of which contains
constituent securities that share at least one attribute in common with one
another and with
the larger segment group. In some cases, a segmented composite unit may be
comprised only
of a larger segmented group and two sub-segments. In other cases, a segmented
composite
unit may be comprised of as many segments as the size and diversity of the
larger segment
will support. A security may be a constituent, in whole or in part, of one or
more larger
segmented composite units. With reference to Fig. 4, a segmented composite
unit can
comprise elements 1205 and 1210.
[113] Stratified Composite Portfolio: As used herein, a stratified composite
portfolio is
defined as comprising at least two stratified composite units wherein the
attributes of the
parents in the composite units represent risk groups such that: 1) parent risk
groups have
differentiable risk profiles; and, 2) the sub-units comprising investment
securities in risk
groups are formed as stratified composite units.
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[114] Segmented Composite Portfolio: As used herein, a segmented composite
portfolio is
defined as comprising at least two segmented composite units wherein the
attributes of the
larger composite units represent risk groups such that: 1) larger risk groups
have
differentiable risk profiles; 2) the sub-segments comprising securities in
risk groups are
formed as segmented composite units.
[115] While there may be other qualifications to be in the parent or larger
grouping of a
stratified or segmented composite unit, respectively, composite unit parents
can satisfy the
condition of sharing a specific common attribute or sets of common attributes
with the
members. A parent grouping of the multiple stratified composite units can
comprise a
stratified composite portfolio defined to create a portfolio of composite
units so that a defined
differential risk is addressed by the composite units that comprise the
stratified composite
portfolio.
[116] Portfolio: A portfolio, as used herein, can be any form or collection of
investment
securities held by an investment company, institution, or individual.
[117] Introduction to Risk
[118] Investments are made with an expectation of appreciation, or return, and
of potential
risk, or variance of these returns. The two measures are linked: at a given
level of liquidity,
the higher the expected risk, the higher the expected return. Stated
differently, all else being
equal, higher levels of risk should be compensated for by higher levels of
return. The
probability of return is linked to the expected variance of outcome for a
given security. The
actual return expected for a security may be tied to many factors including
market conditions,
a given supply of investment capital, or an expectation of inflation or
deflation. For example,
identifying that a company is in the semiconductor business is a
differentiable risk.
Furthermore, the type of semiconductor (e.g., storage, processing, linking) is
important, as
are the raw materials required and the identities of the customers.
[119] Securities vary in their return characteristics and expectations.
Certain types of
securities represent a specific ownership position in a specific company. Each
type, such as a
bond, an equity instrument, or a derivative, has its own specific ownership
and investment
characteristics. The expected return from a security is based on the type of
security and its
characteristics and the underlying performance of the associated entity
relative to the
ownership represented by the security. For any security, the expected return
and the actual
return may be materially different. Theory and empirical results alike
illustrate that
26
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
divergence at the security level is substantially higher for equities than
relatively safe fixed-
income instruments such as investment-grade government bonds.
[120] An investment security's expected rate of return (and volatility)
depends on factors
including both market forces and forces tied to the specific investment
security and its
underlying properties. The former forces are systematic and impact broad
classes of
securities. The latter are specific and unique to each specific investment
security, being tied
to the attributes of each specific investment security or groups of
securities. The variance of
investment security returns that are tied to the latter are tied to attributes
of the specific
securities which are shared in numerous segments of heterogeneous populations.
[121] Risk can be associated with the qualities or attributes of the entity
with which the
security is associated. The changes in fortunes or even bankruptcy of a
specific business are
related to the functional attributes of the business itself These include any
number of factors
including the business, its operations, its products, its customers, its
customer's customer, the
availability of supplies, the strength of their suppliers or the specific
assets or liabilities of the
business. Events related to any one of these things or any combination of
these things can
cause the fortunes of a business to change and, in so doing, change the
expected return of a
business associated with a security.
[122] In addition to an individual company, a portfolio of securities can be
impacted by
these non-systematic risks if the portfolio is over-exposed or over-
concentrated in a specific
non-systematic risk. One of the principal reasons for having a portfolio is to
reduce this
exposure to non-systematic risk by spreading it out over a number of
investments with unique
or disparate non-systematic risks such that no one non-systematic risk will
materially change
the fortunes or expected return of the overall portfolio. This strategy is
relatively easier for
an individual investor who can diversify a portfolio over a relatively small
number of
individual securities in relatively small amounts. However, this strategy has
proven elusive
for large-scale investors such as pension funds or endowments that have
billions of dollars (or
dollar equivalents) to invest. Those large-scale investors must invest in
hundreds or
thousands of securities at any given point in time representing billions of
dollars of value.
For investors with that scale of investment, minimizing the impact of non-
systemic risk
factors in a portfolio has proven very difficult; they tend to overweight in
large industry
bubbles and are negatively affected by repeated technology or commodity
bubbles and
continual over-weighting in large bankruptcies or large downgraded classes of
financial
27
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
instruments such as mortgage-backed securities or sovereign debt. The
invention disclosed
herein provides a method for portfolio managers to systematically control for
these non-
systematic portfolio risks that disproportionately and negatively impact large-
scale portfolios.
[123] Functional Attributes
[124] Functional attributes can be used in the multi-attribute weighting
scheme described
herein. The systems described herein can operate by assigning one or more
attributes to
companies associated with an investment security. The methods described herein
can be
implemented on a computing device to group a portfolio of investment
securities into subsets
using the functional attributes related to their associated companies,
commodities, assets, or
liabilities. These attributes can be used as markers for the specific risks
associated with
events such as bankruptcy or market crashes. These attributes enable a
portfolio manager to
stratify, segment, or sub-divide a portfolio into groups according to
attributes, where each
group represents a specific attribute-related risk. When constructed in a
stratified form, the
children of these parent groups have both unique risks between groups and
share common
risks with their parent.
[125] After stratifying or segmenting a portfolio, weights can be assigned to
the units and a
plan to reconstitute the weightings on a systematic basis can be executed. In
this way, a
portfolio manager can understand and manage the specific risks in the
portfolio.
Additionally, risk levels can be engineered by arbitrarily setting weights for
the stratified
units. In some embodiments, the manager can determine the desired risk at the
beginning of
the process, using these to form a multi-level hierarchy of distinct groups
and sub-groups, and
then weighting the groups according to a desired risk outcome. In other
embodiments, the
groups are used to form non-hierarchical segments, clusters, or groupings and
then weighted
according to a desired risk outcome.
[126] The methods described herein enable the calculation and implementation
of
weighting schemes for portfolios and their constituent securities, each of
which have specific
properties that are different from those of uncontrolled portfolios of the
same securities based
on security or group-specific attributes. As described in more detail below,
the invention uses
a set of security-specific functional attributes that are syntactically and
semantically related to
constituents to reduce the portfolio-level effects of the randomness of
individual security
returns by building portfolios of securities that reduce the impact of the
risks associated with
functional attributes. It does so by stratifying and segmenting the attributes
and their risks in a
28
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
controlled manner over a controlled portfolio of population groupings,
representing
groupings defined by common attributes and groupings containing specific
securities that
share the attributes associated with the grouping.
[127] Stratification and Segmentation
[128] To control for non-systematic risk, a portfolio manager must control for
the specific
set of business risks that exist in any portfolio. These risks could be, among
other things,
company-related, industry-related, product-related, customer-related, or
supplier-related. The
larger a portfolio becomes, the more difficult it is for a portfolio manager
to understand its
exposure to specific non-systematic risks. The methods of risk group
stratification described
herein reduces the negative impact of attribute-specific volatility on the
portfolio as a whole.
[129] The systems described herein can be used to create a stratified
architecture or
segmented sets of specific risk groups, allocating the securities in a
portfolio across these
stratified or segmented risk groups and selecting the desired exposure to the
risk groups by
applying calculated or user-provided weights for identified non-systematic
risks. Thus,
stratification or segmentation can be used to systematically control exposure
to non-
systematic risks. These exposures can then be managed over time by creating
rebalancing
rules that reset on an appropriate periodic schedule a portfolio's exposure to
these identified
non-systematic risks. In this way, a large-scale securities portfolio's
exposure to a set of non-
systematic risks can be systematically determined and managed.
[130] The systems can include a programmable coordinate-guided system to
produce
computer-generated risk groups and programmable assembly of computer-generated
risk
groups into computer-generated portfolios of these risk groups each containing
securities that
match the attributes of the specific group.
[131] Economic entities with one or more common functional attributes
correlate with
events that are associated with that attribute or set of attributes. The
measure of correlation
will vary by the level of importance of that attribute in a specific business.
For example, if all
network equipment companies share the same customers, the loss of a major
customer like
Cisco, a giant network company, will impact all the companies. The impact,
however, will
be greater if Cisco is the company's sole customer than if Cisco is less than
5% of a
company's business. In this way, grouping companies in risk groups that are
defined by
attributes provides a method for portfolio managers to organize, segment, or
stratify securities
in groups that correlate with specific attribute-related events. In addition,
most attributes are,
29
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
in turn, part of larger attribute groups. When the large telecommunications
company Norte'
went bankrupt, all the companies that shared it as a customer were also part
of a network
equipment group which in turn was part of a communication equipment group
which in turn
was part of a larger digital technology group, all of which were exposed to
the bankruptcy.
In this way, using functional attributes enables a portfolio manager to group
securities by
both broad and narrow categories and by the importance of these categories in
determining
the performance of individual securities.
[132] Endogenous economic models characterize functional-attributes, which
represent risk-
related properties, qualities, or characteristics. Coding for these attributes
in a coordinate-
based or ordered tagging system enables a computer to associate tags with
specific risks and
generate company groupings that share these attributes. These risk-based
computer-generated
groupings may be tested, as a non-limiting example, for correlations with
their constituent
companies, with other groups, other tags, or other individual companies or
securities. In an
iterative process, a computer can use the tags in this way to test and
validate the statistical
importance of different computer-generated groupings or individual tags used
to build
computer-generated risk-controlled portfolios of securities that have unique
risk
characteristics derived from the computer-generated groupings. Further, the
computerized
system described herein can be used to generate an assembly of groupings,
including, as a
non-limiting example, a risk-stratified portfolio consisting of stratified
groupings of statistical
control groups.
[133] The process of stratification or segmentation can include dividing a
population into
subsets (called strata or segments) within which one or more investment
securities scan be
placed. Stratification and segmentation can be used in the statistical
management of the
portfolio, as they are used to divide a population into parts or subsets. The
creation of
defined subsets which are assigned defined proportions enables the creation of
controls to
population outcomes through statistical methods.
[134] A properly stratified or segmented population can be termed a control
group because
its constituents and the weights of the subsets are defined and can be tested.
In any
heterogeneous population, there tends to exist random variance wherein a
subset of the
population has different characteristics, properties, or qualities than the
population as a
whole. The impact of these divergent sub-populations can be mitigated by
grouping the
population into sub-populations that are expected to behave differently and
then ensuring that
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
some of each sub-population is used in studying the population as a whole. As
an example, if
one were studying the output of workers, one might find that workers on Monday
morning
were less efficient than the entire rest of the week. However, if one did a
random sample of
20 days worked during a year, one might randomly receive a sample set that was
abnormally
biased toward Mondays. This would not be representative of the workers as the
dataset was
skewed to the one period when workers were least efficient. In an effort to
eliminate this
bias, one might stratify the population set across five subsets consisting of
one subset for each
day of the week. Random sampling would entail assigning each subset an equal
number of
worker days so that the entire sample consisted of five subsets, each with an
equal number of
example days. In this way, stratification can limit biases in a sample set and
increase the
probability of a representative outcome.
[135] Stratification provides controls that can: 1) ensure an unbiased sample
set that is
representative of the entire population; or, 2) ensure a specific exposure to
increase the
likelihood of an outcome that is desired but not necessarily representative of
the underlying
population. An example of the former is in clinical trials or experiments in
the social
sciences. In those cases, the experimenter is attempting to form a
representative sample set
against which assumptions can be varied to investigate how they impact the
controlled
population. An example of the latter is in risk management, where different
population
subsets are designed to be relatively uncorrelated and have highly divergent
occurrences or
variations. In that case, the statistician may want to weight the sample set
towards a specific
subclass, such as subsets that have relatively higher or lower volatility. In
both cases,
stratification enables the statistician to build sample sets with relatively
predictable outcomes
based on the type of stratification model being implemented. The strata
generally are formed
based on members' shared attributes or characteristics. These attributes could
be based on
physically identifiable attributes such as color of hair, skin or eyes, right-
handedness or left-
handedness. In addition, the attributes could be based on relative
quantitative metrics of a
population, such as size, speed, or age of a population.
[136] In the context of investment securities, the value of an investment
security can be
directly or indirectly related to: 1) the type of assets, liabilities, or
operations that are directly
or indirectly associated with the security, and/or 2) the specific functional
attributes
associated with the assets, liabilities, inputs, outputs, products, or
operations that are directly
or indirectly associated with the security.
31
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[137] The aggregate expected return of a composite portfolio created using the
methods
described herein can be determined from the expected returns of each
individual security and
its weight in the composite portfolio. The aggregate expected volatility of
the composite
portfolio can be determined from the expected volatility and weight of
individual investment
securities and the pairwise correlations of these individual investment
securities with one
another. Because of this, the overall volatility can be controlled, and
reduced, by
stratification or segmentation of the portfolio into groups that have
relatively high intra-group
correlation and relatively lower inter-group correlation.
[138] While quantitative values associated with securities are likely to
exhibit significant
astationarity, qualitative attributes are likely to persist over time, driving
performance
characteristics with consistency and facilitating portfolio management and
index construction
at scale. The data systems described herein, which enable the syntactic and
functional tagging
of hundreds of thousands of securities and the dynamic segmentation and
stratification of
large sets of securities by their associated attributes, are instrumental in
enabling this process.
As a non-limiting example, by dividing the securities into correlation
clusters, i.e., groupings
formed based on attributes that correspond to risks, volatility can be
controlled.
[139] Syntactic Attributes
[140] The attributes described above can be represented in a syntax which
defines the
structure of the composite units and composite portfolios. The structures can
be defined by
the use of syntax and architectural positions or coordinates, including the
identification of
attributes related to data entities that are associated with syntactic
positions. Syntactic tags
can have relational attributes that enable syntactic positions to be related
to each other.
[141] As used herein, in some embodiments, a syntax may comprise a set of
rules. A
syntactic position can be defined as a valid position based on the set of
rules. As a non-
limiting example, a syntax may be represented in coordinate space in an
arbitrary number of
dimensions.
[142] A symbol in a database corresponds to a data entity. In some
embodiments, a
syntactic tag associates a symbol and a rule, where the symbols are
constituents of a lexicon,
and the symbols can be combined to form valid expressions according to
principles of the
syntax. A syntactic tag associates the data entity marked by a symbol to the
other data
entities based on the syntax-established set of rules. In some embodiments,
the process of
syntactic tagging provides a means for relating domain-specific information.
It takes
32
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
information in a domain and tags it with rules that relate it in the domain.
Syntactic tags can
be dynamic.
[143] In some embodiments, a syntax can be used to evaluate the validity of
expressions in
a system. A symbol in a database can be used to mark a data entity. A
syntactic tag can be
used to mark the association between a symbol and a mechanism for evaluating
the validity
of expressions. The tags may be of multiple types, including syntactic
attribute tags which
ascribe relationships between symbols and rules that characterize attributes.
In some
embodiments, a syntactic tag associates the data entity marked by a symbol to
the other data
entities based on the syntax-established set of principles. As a non-limiting
example, this
process of syntactic tagging provides a means for relating information within
a domain or a
subset thereof, or across domains.
[144] Syntactic tags can have some or all of the following properties:
[145] Expressions can be combinations of labels for tags. In some embodiments,
such
expressions can conform to a syntax expressible in BNF (Backus Normal Form or
Backus¨
Naur Form) notation or an equivalent meta-notation.
[146] Any valid expression or sub-expression consisting of more than one
element of the
syntax, can form a locus.
[147] Any element of the syntax that has a range of potential values describes
a dimension
in a discrete multidimensional space consisting of the dimensions associated
with all such
elements.
[148] Any expression or sub-expression of the syntax, containing elements
which have a
range of potential values, may be stratified, in which case that expression or
sub-expression
describes a dimension which consists of regions and successive sub-regions
within the multi-
dimensional space. As a default, elements of syntax which are designated as
stratified are
interpreted from left to right according to their position within the
expression, as successive
levels from top to bottom within the architecture.
[149] Syntax can represent coordinates that provide successive specialization;
the degree of
specialization grows with the depth of the architecture. The syntax can also
provide step-wise
serialization at each level; the degree of serialization grows with the number
of elements at
each level.
33
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[150] In some embodiments, at each level of specialization and/or degree of
serialization,
the syntax elements share a proximate syntactic position with both:
[151] a) their parent in the architecture; and
[152] b) their siblings in analogous positions across different architectures
in the same
syntax.
[153] Syntax elements may be considered to have a proximate syntactic position
if they are
relatively close to other elements based on either their specialization or
serial positions.
These relationships allow for comparison of values across syntactic positions.
This property
supports applications including but not limited to the complex structures,
population sorting,
autoclassification, and integration with prior art temporal and spatial
classification systems.
[154] A functional information system (FIS) can be implemented as a database
system
which utilizes syntactic tagging and the related concept of a locus, as a
logical model for
organizing data about a domain. A basic implementation of the FIS can be
achieved by
having a store of the syntactic terms of the FIS to augment the store of data
entities in the
domain. Each data entity can have a reference to its location in the FIS.
These table
references enable searching for all data-entities in a specific position as
well as searching for
the position of any data entity in the system.
[155] Syntactic tags are assigned to structured or unstructured data, either
manually or via
an automated process and can be associated with a unique identifier for each
data entity.
When sets of data entities are associated with abounded, well-known range of
objects or
entities, then a lexicon containing standardized identifiers may optionally be
used to facilitate
the assignment of identifiers to data entities.
[156] As a non-limiting example, syntactic tags can be used to represent the
syntactic
components of a domain-specific data entity. They can be used for recording
and storing
information that indicates to a user how specific data entities relate to each
other and/or to the
specific domain. The tags can be used to determine which data entities are
similar and/or
why they are different and or to what degree they are different.
[157] The domain-specific rules described herein can be used to characterize
the syntactic
components of data entities in a domain and populate sets of domain-specific
syntactic tags.
They can be assigned to any domain-specific data entity associated with a
domain-specific
syntactic position. Once assigned, stored and retrievable, the data entity can
now be related
34
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
with any other data entity that shares any value on its syntactic tag. It can
be used for
grouping of information based on, for example, broad values or very specific
values. If the
values are broad, it provides the ability to create ever-smaller sub-sets
within the context of
the broad set. If other domains share the same syntax, the tags can be used to
compare data
entities in one domain to data entities in other domains based on shared
syntax.
[158] The rules of syntax can be based on an arbitrary number of factors. As
non-limiting
examples, they could be based on common temporal order, spatial order,
anatomical,
morphological, physiological, or mechanical order. The rules could be areas
specialized to a
specific domain such as the order of its influences or of its origins. The
rules could be
experimental and the validity of the rules could be tested using syntactic
tags. In each case,
the knowledge influenced by some ordering principle has a syntax that provides
the rules for
the ordering. Once recorded, stored, and retrievable, the process of relating
data entities
based on syntactic tags can be based on established rules defining how
different data entities
relate. This system can be applied to any domain and any syntax. In so doing,
it provides a
tool to add dimensionality to information from any field. It can also provide
a procedure for
converting a legacy system from any field into this framework by applying
syntactic tags to
the legacy codes.
[159] Syntactic positions in the system have specific attributes that are
associated with the
rules of the syntax. For example, if a domain-specific syntax is a temporally-
based syntax,
the attributes will be temporally related; if it is a spatially-based syntax
the attributes will be
spatially related; or, if the syntax is mechanically-based, the attributes
will be mechanically
related. If the syntax is sequential, the attributes will be sequentially
related. If the syntax is
nested, the attributes will be related to the rules of nesting.
[160] In some embodiments, to create syntactic tags, a domain is defined, then
a domain-
specific syntax is defined. In one embodiment, the system can be configured so
that the
specific rules of the domain-specific syntax are fully represented in domain-
specific syntactic
tags.
[161] Syntactic tagging links data entities with shared attributes by
assigning each data
entity to an element in the set of common syntactic tags. The syntactic tags
associate data
entities with the other data entities in a domain according to their syntactic
associations.
Thus, they inherently group and/or cluster all data entities that share
syntactic tags.
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[162] In some embodiments, syntactic tags can be assigned to data entities
which have one
or more attributes in common, or the same or similar meaning, in a context of
interest for the
domain to which the FIS is applied. By tagging data entities with data-entity-
type tags, the
system can operate on multiple different kinds of data within a domain or data
set. For
example, data for products or markets can be added to company data. This
function can be
used in connection with flagging functions, described below, to indicate that
certain tags may
be required only for specific data-types.
[163] Syntactic tags can be used to express:
[164] (1) successive specialization, whereby all data entities that share the
same tag at a
higher level also share certain common characteristics or meanings within the
domain; and
the ordering of such labels within a level is a matter of tag assignment
convention, or is
arbitrary; and/or
[165] (2) a sequential process whereby all data entities that share the same
tag at the next
higher level also share the common characteristic that they are successive
steps the same
sequential process of the domain, at the same level of process-detail; and the
ordering of such
labels within the category directly reflects the sequence of steps.
[166] In some embodiments, the complete enumeration of the valid syntactic
tags provides a
complete pre-existing model for the structures of interest in the domain to
which the FIS
model is applied, regardless of whether any data is actually tagged with any
given label.
[167] Syntactic tags for stratified composite units can be combined to form
expressions.
Such expressions can conform to a syntax expressible in BNF notation or an
equivalent meta-
notation. Any expression or sub-expression of the syntax, containing elements
which have a
range of potential values, may be stratified, in which case that expression or
sub-expression
describes a dimension which includes regions and successive sub-regions within
the multi-
dimensional space.
[168] Syntactic elements may be considered to be proximate if they are
relatively close to
other elements based on either their symbolic representation or serial or
complementary
positions. These relationships allow for the comparison of values across
syntactic positions.
[169] Syntactic tagging of the attributes links data entities with shared
attributes by
assigning data entities to an element in the set of common syntactic tags. The
syntactic tags
associate data entities with the other data entities according to their
syntactic associations.
36
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
Thus, they may group or cluster data entities that share syntactic tags. In
some cases,
syntactic tags can be used to create a normative model for a portfolio,
discussed in more
detail below.
[170] The systems described herein can be used in combination with a barcode
that
identifies a multitude of business attributes. The system can assign this
standardized barcode
with functional attributes and syntactic tags to securities in a portfolio.
Based on this barcode
of attributes, specific non-systematic risk exposures that exist in a
portfolio can be identified.
Once identified, the method can be used to control for these non-systematic
risks by limiting
a portfolio's exposure to these risks.
[171] An example representation of an architecture developed from syntactic
tags is
illustrated in Figs. 8A and 8B. A graphical representation is illustrated in
Fig. 9.
[172] Portfolio Architecture Creation
[173] Constructing large-scale portfolios of securities is challenging for
numerous reasons.
It is difficult without both a reliable and validated system of attributes as
well as a
stratification or segmentation system that uses a stratified composite
architecture or
segmented composite units to control for the large number of functional
attributes that
influence performance at the security, group, and portfolio level.
Independently and together,
the systems and methods described herein enable the engineering and management
of risk
exposure on a large-scale basis.
[174] An engineered composite of investment securities is a group of
securities that are
engineered (or selected) to possess a different risk/return profile than an
uncontrolled
grouping from the population of underlying securities or the underlying risk
groupings that
are used to construct the composite.
[175] Stratified or segmented composite portfolios comprising investment
securities can be
based on a dynamic combination of entities of a proximate class to produce a
new unit
consisting of a part of each of the constituents being combined to create a
new entity that has
different properties from the underlying constituents taken separately.
Dynamic properties
mean that the properties of investment securities vary and change over time.
Investment
composites can be configured to account for this dynamic nature in order to
create reliable
composites that substantially maintain their properties over time.
37
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[176] A method for building a stratified composite portfolio using a syntax
for investment
securities can include the following steps: 1) grouping investment securities
with common
risk attributes; 2) stratifying or segmenting the grouped investment
securities into sub-groups
that are a) associated with different risks, while b) still associated with
the risk characteristics
of the group in which they are contained.
[177] In one embodiment, a composite portfolio can include an identification
of multiple
securities and their associated weights. As a non-limiting example, the
identifications and
weights can be executed using a computerized process according to the example
method
illustrated in Fig. 1. As illustrated in Fig. 1, the method can first generate
a stratified
portfolio architecture (1125) and then a resultant list of investment
securities and weights
(1150). In an initial step, a stratification module (1105) can receive as
inputs investment
security-related attributes (1120) and an architecture of attribute rules
(1122), both of which
can be stored on one or more computerized data storage devices. As non-
limiting examples,
the investment security attributes can be selected from those examples
provided above. Other
attributes and types of attributes can be used.
[178] The attribute rules can be provided by the portfolio architecture, as
described above.
The architecture can be used to define or evaluate relationships among
attributes, tags, values,
and the investment securities associated with the attributes.
[179] The stratification module (1105) can also include a selection submodule
(1110) to
receive, as input, a selection from a user of attributes (1120). In some
embodiments, the
functional attributes characterizing the economic entities enables the
construction of
portfolios from securities associated with those entities. As a non-limiting
example, a syntax
permitting the evaluation of expressions characterizing economic entities is
illustrated in
Figs. 8A-8B. In other embodiments, the syntax can be adapted to attributes
selected by the
user. In other embodiments, the user can be provided with an interface for
creating new
structures (1121) which are then inputted to the stratification module (1105).
[180] In some embodiments, a structure can be created from a Boolean statement
in the
form of 'attribute' operator"value' that may return true or false for an
entity or its associated
investment security based on its attributes. In other embodiments, a structure
can be created
a Boolean expression that combines (via Boolean operators) one or more
statements. The
lines in Fig. 9 illustrate examples.
38
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[181] In some embodiments, a architecture can be defined as a relationship
among a set of
structures that defines the portfolio segments, under the constraint that any
entity or its
investment security that fails at one node in the structures will not be
passed through the rules
of any of that parent's children. The stratification submodule (1115) can be
configured to
create a stratified portfolio architecture (1125) based on the set of
structures (1122),
investment security attributes (1120) (optional at this stage), input
regarding the creation and
selection of structures (1121), or a listing or other identification of
investment securities
(1131). The stratified portfolio architecture (1125) can then be
electronically represented and
stored on a computerized data storage device.
[182] A structure can be derived from one or more statements that filter
entities and
investment securities based on attributes. As a non-limiting example, a
stratified structure
can be used to define a relationship among structures. Any company that is
excluded from a
top level will also be excluded from lower groups. The multiple attribute
system described
herein can be configured by varying the population in any parent or child by
varying one (or
more) of the attributes defining that parent or child. The ordered rules can
also be expressed
as a graph or network, which can be configured by enabling the population to
be dynamically
ordered based on functional attributes defined by the computerized system, the
user, or a
combination thereof
[183] Example graphical and textual representations of a resultant stratified
portfolio
architecture are illustrated in Figs. 3 and 4. Fig. 3 illustrates example
attributes and their
syntax. The attribute-based rules illustrated in Fig. 3 are graphically
presented in Fig. 4. The
rules illustrated in Fig. 3 describe a top level composed of two groups having
enterprise loci
of real estate (1; 1205) and equipment materials manufacturers (2; 1210). The
rules in Fig. 3
further describe enterprise loci of real estate developers (1.A; 1215), real
estate operators
(1.B; 1220), REITs/real estate lessors (1.C; 1225), manufacturers of materials
for
information-processing equipment (2.A; 1230), and manufacturers of materials
for non-
information-processing equipment (2.B; 1235). These enterprise loci are
illustrated at level
two of the stratified architecture. The rules in Fig. 4 include several third-
level relationships.
The third-level defines relationships for consumer real estate developers
(1.A.i; 1240),
industrial real estate developers (1.A.ii; 1245) under real estate developers
(1.A; 1215); North
American real estate operators (1.B.i; 1250), European real estate operators
(1.B.i; 1255), and
Asian real estate operators (1.B.i; 1260) under real estate operators (1.B;
1220); and low-
leverage REITs (1.C.i; 1265) and leveraged REITs (1.C.ii; 1270) under
REITs/real estate
39
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
lessors (1.C; 1225). Further relationships are illustrated under groups (2.A;
1230) and (2.B;
1235), but are not further described here.
[184] Numerous attributes may be used to create a portfolio architecture. The
portfolio
architecture can include a nested structure of groups. As a non-limiting
example, in some
instances, these groups can be formed by referencing the attributes which are
common to all
entities in the universe, such that at each level, every element of the
universe is in exactly one
group. In some embodiments, these groups may be sub-divided into an arbitrary
number of
child sub-groups ¨ and this number need not be the same for each of the
original parent
groups ¨ and this sub-division process can be carried out an arbitrary number
of times, each
time adding a level to the architecture in a "top-down" manner. In some
embodiments,
stratified composite units are used to build larger stratified composite
units, creating a
structure in a "bottom-up" manner. In some embodiments, a combination of "top-
down" and
"bottom-up" approaches may be used. In other embodiments, existing economic
and financial
classification schemes may reconfigured using syntactic tagging to make them
relational and
dynamic, and be partially or wholly used in the portfolio architecture in
combination with any
or all of the universe selection, weighting, reweighting, and rebalancing
schemes described
herein. Regardless of the construction method, the resultant portfolio
architecture (1125) can
comprise an electronic representation of a set of attributes arranged, as non-
limiting
examples, in graphical, segmented, stratified, or network form, according to
the defined
attribute rules.
[185] Weighting of Investment Securities
[186] A stratified or segmented composite portfolio can be constructed of one
or more
stratified or segmented composites that maintain defined risk exposures by
weighting the
constituents of the stratified or segmented portfolio accordingly.
[187] The stratification or segmentation described herein can be adjusted in
various ways to
enable a user to control the population of investment securities and thus the
outcomes that
arise from events associated with a population of investment securities.
Portfolios can be
adjusted, and resulting performance metrics can be engineered, based on
changes made to
any or all of: 1) the population of investment securities; 2) how the
population of investment
securities is stratified or segmented (the portfolio architecture); and, 3)
how the stratification
or segmented units are weighted within the architecture, graph, or network.
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[188] Once the portfolio architecture has been determined, weights can be
determined for
the securities. As a non-limiting example, a weighting function can be any
function that, for
a specific group in a stratified portfolio architecture, returns a value
between -1 and 1
indicating the weight associated with that group relative to its siblings in
the portfolio
architecture. In some embodiments, the absolute value of a weight may exceed
1. As non-
limiting examples, negative weights can be implemented by short selling, and
weights whose
absolute value exceeds 1 can be facilitated through leverage. In some
embodiments, the sum
of the weighting function for all the siblings or composites at each level or
unit can be equal
to 1.
[189] In some embodiments, a security's weight is only a function of its
position in the
architecture. As a non-limiting example, among strata, weights may be divided
evenly
between all of the children of a given parent group. That is, if the first
level contained 10
groups, each would be given a weight of 10%. If one of these groups contained
4 sub-groups,
each would be given a weight 25% of its parent group, for a resultant weight
of 25%*10% =
2.5%; while if a different top-level group had 5 child groups, each child
would weigh
20%*10% = 2%. This process can be repeated for each level, eventually yielding
a weight
for each bottom-level group. A similar process can be applied to securities
within each
bottom-level group, yielding weights for each security in the universe.
[190] In some embodiments, the weighting algorithm can be executed by a
computer, as
follows:
class PortfolioGroup
# Returns a list of the portfolio groups
# at the same level as this portfolio group
def siblings
end
# returns a parent of this portfolio group.
# if this portfolio group does not have a
# parent, it returns undefined.
def parent
41
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
end
# returns the weight that should be associated
# with this portfolio group.
def weight
num of siblings = self siblings, count
if parent.is defined?
parent weight = self parent.weight
else
parent weight = 100
end
return 1/num of siblings * parent weight
end
end
[191] In other embodiments, the weight of any group may be derived from the
incidence
attributes of the companies in that group. As a non-limiting example, groups
(formed using
any of the attributes) may be weighted by a function of one or more of the
attributes common
to securities in the universe. As a non-limiting example, groups may be
weighted within their
parent group proportional to the total debt of all securities in the group. In
some
embodiments, the function depends on a single attribute. In other embodiments,
the function
depends on a plurality of attributes. In some embodiments, the same function
is used to
weight every group in the architecture. In other embodiments, different
functions may be
used to weight different groups in the architecture. In some embodiments, the
weighting can
be executed by a computer, as follows:
class PortfolioGroup
# Returns a list of the portfolio groups
# at the same level as this portfolio group
def siblings
end
# returns a parent of this portfolio group.
42
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
# if this portfolio group does not have a
# parent, it returns undefined.
def parent
end
# A function that for a specific group in
# a stratified portfolio architecture returns
# a value between 0 and 1 indicating the weight
# associated with that group relative to its
# siblings in the portfolio architecture.
# The sum of the weighting function for the
# siblings at each level equals 1.
def weighting
end
# returns the weight that should be associated
# with this portfolio group.
def weight()
if parent.is defined?
parent weight = self parent.weight
else
parent weight = 100
end
return weighting * parent weight
end
end
[192] With reference to the example of Fig. 1, computerized weighting module
(1130)
receives the portfolio architecture (1125). As illustrated in Fig. 2, the
weighting module can
also be configured to receive identification of investment securities (1131),
and identification
43
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
of attributes (1132) associated with the securities. The weighting module can
then generate a
list of securities and associated weights (1150). The weighting module is
illustrated in
further detail in Fig. 6. As illustrated in Fig. 6, the system can receive a
selection and/or
identification of the investment securities to be weighted (1305). The
investment securities to
be weighted could be positioned at any point or points in the architecture
described above.
Weightings for individual securities and groups of securities can then be
calculated for the
current level or segment (1310). In some embodiments, the calculation can
start at the top
stratum. At the current level, the weighting scheme and rules (1315) for that
level are
identified. A weighting coefficient can be calculated by dividing the
outstanding proportion
of weight by n, the number of investment securities or groups of securities
(1320). As a non-
limiting example, with reference to Fig. 4, the top-level weighting may be
calculated to be
50% to Group 1 and 50% to Group 2. At the second level, Groups 1A ¨ 1C may be
weighted
at .50*.333 = 0.167 or 16.7% each.
[193] Before or after calculation of the weightings, any positive or negative
weighting
biases may be applied (1325). Biases can be applied by arithmetic or other
operations on the
weightings. In some embodiments, any biases that are applied to one group or
investment
security require a corresponding opposite bias to be applied elsewhere in the
same group or in
a peer group at the same level. If the bottom level has been reached and
completed, the
weighting process may terminate. Otherwise, the process may continue at the
next level.
[194] The electronic representation of the weighted investment securities can
then be input
as instructions to, as non-limiting examples, an exchange traded fund (ETF) or
another
financial instrument such as a hedge fund, mutual fund, limited partnership or
another
investment vehicle.
[195] In alternative embodiments, the steps of the method for stratification,
segmentation,
and weighting can be reordered. For example, the list of investment securities
could be
introduced anywhere in the portfolio engineering process. Investment
securities and/or a
reconstitution process could be chosen before stratification or segmentation
to create
exposure to a particular universe. An architecture, weighting scheme, or
rebalancing scheme
could be selected or chosen before or after choosing the investment
securities.
[196] Alternative orderings and variations of the steps for creating the
portfolio of
investment securities described above are possible. For example, with
reference to Fig. 1, the
identification of investment securities (1131) can be provided to the
stratification module
44
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
(1105). In that arrangement, the stratification submodule can generate the
stratified portfolio
architecture of investment securities (1125) that is then input to weighting
module (1130).
[197] In some embodiments, universe identification, group selection, and
performance
characteristics can be combined into one module. In other embodiments, frames
representing
queries, structures, and outputs can be combined into one module. The
portfolio and its
constituent groups, composites, and/or securities may be represented, as non-
limiting
examples, in stratified, segmented, networked, or graphical format, or in a
daisy chart. In
some embodiments, the outputs may be selected from a chart, geographic map,
tree map,
microarray, or table.
[198] Reconstituting and Re-weighting
[199] Additionally, some embodiments can include reconstituting the designated
segment or
group weights on a periodic basis to maintain the desired risk exposures. A
stratified or
segmented portfolio can be comprised of one or more stratified or segmented
composite
units, respectively, that maintain defined risk exposures by weighting the
constituents
accordingly and reconstituting the designated weights on a periodic basis to
maintain the
desired risk exposures. With reference to the embodiments illustrated in Figs.
1, 2 and 5, the
steps illustrated can be performed at any arbitrary point to create a re-
weighted portfolio
based on modified inputs, such as modified weighting rules. With reference to
Fig. 5, in
other embodiments, the re-weighting can be provided by a separate re-weighting
module
(1155). The re-weighting module (1155) receives a list of target exposures
assigned to
portfolio groups, composites, or constituents (1151). The re-weighting module
then selects
new investment securities for inclusion in the stratified composite portfolio.
[200] Composite Portfolio Scoring
[201] Using methods described herein, a score can be calculated for a
composite portfolio.
The score can be a characteristic of the portfolio and can be used in multiple
contexts. In
some embodiments, the target score can be a number that the portfolio is
engineered to reach.
In other embodiments, the target score can be a set of attributes that an
investor would like
the portfolio to have. The portfolio score can be a value or vector of values
calculated from
the portfolio which can be compared with a target score an investor has for
the portfolio. The
target score can be a theoretical or estimated value.
[202] A target score can be used as a way to optimize a portfolio. The
investor can pick the
target score and the system can then be used to build a stratified composite
portfolio
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
optimized for that score. Alternatively, a target score can be used to build a
portfolio that
reflects the performance of the underlying population. That is, the target
score can measure
expected population performance, and the stratified or segmented composite can
be used to
measure actual population performance. Given a weighted list of securities of
a portfolio and
a target score, the score for the portfolio may be calculated based on derived
attributes of a
portfolio.
[203] The target score can represent an estimate of expected or targeted
portfolio
performance. The target score can be achieved by measuring the performance of,
as non-
limiting examples, individual companies, randomly sampled individual
companies,
stratification units, segments, and/or composites.
[204] The target score can also be identified as the target score that the
investor seeks as
part of the investment objective. Here, the investor may want to use a
stratified or segmented
composite to reach a predetermined target score. By building groups based on
common
attributes, risk groups can be formed. These risk groups may then be weighted
appropriately
to achieve the target score, resulting in a portfolio with known biases.
[205] In some embodiments, a stratified or segmented composite portfolio may
be
engineered to meet a user-defined target score. As non-limiting examples, a
target score
could include any or all of: (a) absolute return goals (e.g., expected rolling
rates), (b)
risk/return measure (e.g., Sharpe ratio, Sortino ratio, or alpha), or (c) risk
goal as measured by
volatility (e.g., downside deviation or beta). In some embodiments, a target
score may be a
one- or multi-dimensional vector of values or elements, such as those examples
provided
above. As a non-limiting example, the target score could be [the actual return
- the risk free
rate] / [the expected return - the risk free rate] where the target score is
greater than or equal
to one.
[206] A method for constructing a stratified composite with a target score,
according to one
embodiment, is described below with reference to Fig. 7. As an initial step,
the user
establishes a population in which to invest by identifying a universe of
investment securities
(7005). The population could be, for example, financial and energy companies
in the U.S.
Next, the universe of securities is filtered (7015). The population of
companies is then
stratified (7020). By this process, they are placed into stratification units,
or groupings based
on common functional or syntactic tags, values, or attributes.
46
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[207] After population stratification or segmentation, the metrics are
identified that will be
used to evaluate the portfolio. The metrics used can depend on the population
that is being
stratified. For example, the metrics used for an investment-grade debt
portfolio may be
expected yield and volatility, while the metrics of an equity portfolio may be
expected risk
and return. Once the metrics have been identified, a target score can be
established (7010).
The target score is the goal that the user would like to see the portfolio
achieve, the goal
being measured by the identified metrics. For example, the target score of an
investment
grade debt portfolio can be an expected yield and expected volatility that an
investor would
like the portfolio to achieve. Example embodiments of the target score are
described below.
[208] Once the target score is set, an engineered composite portfolio can be
created (7020).
Composites can be combinations of two or more stratification units which can
be engineered
to reach the target score. Composites can be engineered by strategically
weighting
stratification units and the companies within the stratification units (7025)
and reweighting
the constituent companies (7030). The weighting and re-weighting process can
include
changing the population's constituents (adding or deleting constituents from
the portfolio that
meet the population criteria).
[209] The composite can be tested against the target score (7035). If the
target score is
accepted, the process can reach completion. If the target score is not
satisfied, then some or
all of various parameters can be adjusted, including 1) the architecture
rules, 2) the weighting
rules, 3) the universe filtered through the structure and weighted, and 4) the
rebalancing/reconstituting policies. The process can be repeated until a
portfolio with a
satisfactory score is created.
[210] A stratified composite can be used to optimize a portfolio. As described
above, an
engineered composite can be constructed to meet a target score. Here, the
target score can be
considered the investment objective. For example, the objective could be to
build a
composite whose return, performance, variance, or other property, quality, or
characteristic
matches what is outlined in the target score.
[211] Therefore, instead of building a portfolio that is most representative
of the underlying
population, a portfolio can be created that strategically weights the lower-
level groupings so
that the portfolio will match most closely its target score. Here, stratifying
or segmenting the
portfolio and building composites enable the identification of risk groups
within a population.
Weights thus can be strategically allocated across them in order to meet the
target score.
47
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[212] In investment securities, the primary concerns for investors are risk,
expected return,
and liquidity. Therefore, in some embodiments, the target score may reflect
the investment
objectives of the portfolio quantified with respect to the portfolio's risk,
expected return, and
liquidity characteristics. The goal in creating investment composites is to
engineer the risk,
return, and liquidity through composite design and weighting of the underlying
constituents.
The engineered investment composites can produce composite scores (calculated
from
combining individual security data impacted by multiple attributes) that
reliably can achieve
theoretical estimates.
[213] Using the methods described herein, composites can be engineered to
improve upon
these functional properties, which can be identified or designed for use in
specific
environments. In categorizing investment securities, composites can be formed
to manage
composite scores. A stratified or segmented composite can be used to achieve a
target score.
Stratification or segmentation allows identified risks to be grouped within a
portfolio.
Therefore, when creating an engineered portfolio that meets a target score,
risks to which the
portfolio will be exposed can be better understood qualitatively and
quantitatively.
[214] Synthetic Conglomerates
[215] The methods described herein may be used, as a non-limiting example, as
a means to
achieve through functional diversification a targeted point on the portfolio
risk-return-
liquidity frontier. The data systems described herein enable the synthesis of
instruments that
achieve the diversification at scale sought by conglomerate managers, holding
companies, or
investors in private equity firms, without incurring the high transaction
costs associated with
private market transactions or significant operating expenses. In some
embodiments, a
synthetic conglomerate is an engineered composite, that, as a non-limiting
example, can be
configured to achieve a certain target score.
[216] As a non-limiting example, the management of the synthetic conglomerate
can be
effectuated in real-time by the data systems described herein, by permitting
the dynamic
aggregation of the financial statements of each of the constituents of large
portfolios and the
calculation and display of their consolidated balance sheets, income
statements, and cash
flow statements. The technologies described herein permit customized
identification and
selection of exposures within large-scale portfolios across functional,
temporal, and
geographic space.
48
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[217] In some embodiments, the preparation of earnings estimates and projected
financial
statements at the portfolio, engineered composite, or synthetic conglomerate
level permit the
establishment of a trackable internal benchmark. This customized portfolio,
engineered
composite, or synthetic conglomerate can be compared to those earnings
estimates or
financial statements to determine whether they met or exceeded projections. As
a result, the
customized portfolio, engineered composite, or synthetic conglomerate can be
compared
reliably to internal projections rather than relying, as other portfolios and
indices are required
to do, on external benchmarks.
[218] In some embodiments, the data systems described herein enable the
creation of
streams of earnings, dividends, and cash flows at the portfolio level that are
more stable,
consistent, and predictable than those at the group level. In other
embodiments, the streams at
the group level will be more stable, consistent, and predictable than those at
the security
level. In other embodiments, the streams at the portfolio level will be more
stable, consistent,
and predictable than those at the security level. In some embodiments, the
engineered
composite or synthetic conglomerate can be considered a benchmark that
delivers more
consistent, stable, and predictable returns that more reliably attain the
rates of risk and
liquidity-adjusted return predicted by financial theory than other
commercially available or
widely held indices or benchmarks.
[219] Portfolio Graph
[220] A graph of a heterogeneous population of securities and their associated
functional
attributes, tags, and/or values may be constructed based on an underlying
functional syntax in
conjunction with semantic tags and attributes, geographical and temporal data,
and associated
measures and metrics. As a non-limiting example, a graph of data entities
representing a
population of investment securities or financial instruments is described
below.
[221] The investment securities or financial instruments are assigned nodes on
the graph; as
non-limiting examples, the data entities may correspond to historical or
current companies,
sectors, products, securities, investments, loans, or components,
aggregations, inputs, or
outputs thereof
[222] The nodes are connected based on the relationships (demarcated by edges)
between
the underlying economic entities, including, but not limited to, those
codified in the
functional syntax, geographic or temporal relationships, or those derived from
proximate
economic relationships in the referent system, including supplier-business,
intermediary-
49
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
seller, work group-department, sector-industry. In some embodiments, the graph
is a directed
graph.
[223] The nodes can be grouped visually based on proximity relationships
derived from the
functional syntax, the semantic tags and attributes, and associated measures
and metrics. As
non-limiting examples, the relationships may be ordered and represented
through spectral
analysis, eigenvector clustering, and k-means clustering.
[224] In some embodiments, the edges may be weighted or colored based on the
extent of
interdependence among the nodes or the categorical relationships they reflect;
as non-limiting
examples, this may be derived from trade, transaction, investment, or
financing data among
the entities and their economic referents, commonality of semantic tags or
attributes,
proximity of geographic or temporal relationships, or from the underlying
functional syntax.
[225] In some embodiments, the nodes may be weighted or colored based on the
size or
scale of the entity, or any of the categories associated with the referent
data; as non-limiting
examples, this may be derived from trade, transaction, investment, financing,
or other capital
markets or accounting-based data associated with the entity, semantic,
syntactic, or functional
tags or attributes associated with the entity, geographic or temporal data
associated with the
entity, or measures and metrics associated with any of the foregoing.
[226] The visual representation of the graph and its component parts may be
derived from
default preferences specified by the system, preferences expressed by one or
more users, or a
combination thereof
[227] The graph may be updated dynamically to reflect changes in the
relationships among
entities, permitting an visual representation of an evolutionary model for a
portfolio.
[228] Portfolio Fields
[229] The model of the system may be represented, as a non-limiting example,
as a field on
which mathematical operations can be performed, which facilitates the study of
the
interactions among the economic entities or the issuers of the investment
securities.
[230] In some embodiments, a set of economic entities E and a structure on
that set S,
comprised of subsets si, 2...n of those economic entities, may be stored. In
some embodiments,
the structure is an element of the power set of E, P (E). A set of attributes
A= lai, 2 n1 may be
mapped to those subsets, based on the set a set of values V=Ivi,2 n1, such
that each each a c
Ais a mapping a: S ¨> V. . In other embodiments, the entities may be non-
economic.
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[231] In some embodiments, the field will be an ordered tuple (E, S, A). In
other
embodiments, the field will be an ordered tuple (E,S,A,V). In other
embodiments, the field
will not be ordered. In some embodiments, the entities can be combined to form
expressions,
or ordered sets which can be evaluated by a syntax. In other embodiments, the
expressions
will be one or more combinations of entities which lack order, or which cannot
be evaluated
by a syntax. In some embodiments, a portfolio, group, sub-group, stratum, or
segment may be
characterized as a field.
[232] As non-limiting examples, the model of the system characterized by the
field may be
syntactic, semantic, visual, qualitative, or quantitative, or some combination
thereof As non-
limiting examples, the field may be represented graphically, hierarchically,
or in clustered or
networked form. In some embodiments, the representation will satisfy the
formal
mathematical properties of a field. In other embodiments, the representation
will not satisfy
the formal mathematical properties of a field.
[233] Investment Returns for Securities
[234] In some embodiments, for any given security s, its return r over a time
period t can be
described as
k ,f ff(a)dwdadi f nm(t)dt
where a1,2...õ are the attributes in a given time period that influence the
return of the security,
W1,2....n are the weights to be assigned to each of those attributes, k is a
constant, and n is a set
of equations modeling stochastic components.
[235] In some embodiments, the model can be used in conjunction with the
mathematical
field representation to map the effects of the attributes on performance
characteristics across
groups of securities. In other embodiments, this return formula can be used
for predictive
modeling, diagnostics, or recommendations. In some embodiments,
nõ(i)di
will be 0; in other embodiments, it will be nonzero.
[236] Investment Statistics for Stratified Composite Portfolios
[237] A portfolio generated according to the methods described herein can be
scored using
modified versions of known statistical indicators, including, as non-limiting
examples, alpha,
beta, and Sharpe and Sortino ratios. A score can be generated based on a
normative stratified
51
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
or composite model portfolio and variations on the normative portfolio. For
example, a
stratified or segmented alpha can be calculated as a risk-adjusted premium to
a score on
normative portfolio. A stratified or segmented beta with respect to a
normalized market can
also be calculated for a stratified or segmented portfolio where the
normalized market is
defined to have a beta of 1.
[238] In some embodiments, normative stratified or segmented betas can be
calculated with
respect to any market portfolio, e.g., as non-limiting examples, stratified or
segmented
composite portfolios of the total market, or a subset thereof For example, the
contextual
subset could be defined, as non-limiting examples, as a sector, industry,
geographic region,
time period, dictionary term, or thesaurus term.
[239] Financial Recommendation Engine
[240] The method described herein can be used to recommend securities,
composites, and
portfolios to users. These recommendations are derived from the securities and
their referent
economic entities' syntactic and empirical relationships; the functional,
syntactic, semantic,
temporal, geographic, financial, or economic tags, attributes, and values
assigned to the
economic entities; the express and revealed preferences of the users of the
database or
software; and the relationships of the users in the network.
[241] The relationships embodied in the syntactic tags, attributes, and values
assigned to the
economic entities enable, as a non-limiting example, an initial default
calculation of
proximity among them. In some embodiments, entities sharing common or
proximate values
or attributes associated with a plurality of tags, loci, or partial or full
sequences thereof may
be proximate within one or more databases used to provide recommendations,
while entities
with a plurality of disparate or divergent values or attributes associated
with a plurality of
tags, loci, or partial or full sequences thereof may be disparate within those
databases.
[242] Proximity may also be derived from the empirical relationships among the
securities
and economic entities, which can be aggregated, stored, and assigned to the
data entities and
their referents. In some embodiments, these may include [supplier-customer],
[investor-
entrepreneur], [impact investor-social enterprise], [intermediary-customer],
[customer-
customer of customer], [lender-borrower], [input-output], [employer-employee],
[company-
department], [general partner-limited partner], [service provider-client],
[department-work
group], [subject-activity-direct object-indirect object], [parent company-
subsidiary], [raw
52
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
material-basic component], [basic component-complex component], or [complex
component-
final product].
[243] These empirical relationships may be weighted, scored, timestamped, or
geotagged,
and stored in one or more databases as a basis for proximity calculations. In
some
embodiments, economic entities sharing numerous recent or heavily weighted
relationships
with one another, or with a common third party, will be proximate within one
or more
databases used to provide recommendations, while economic entities without
common
relationships, or whose relationships are purely historical, will be distant
within those
databases.
[244] Proximity relationships may also be derived from the non-syntactic
proprietary tags,
attributes, and values assigned to the economic entities and securities; as
non-limiting
examples, these tags may be functional or semantic. In some embodiments, these
tags,
attributes, and values may include [raw material], [basic component], [complex
component],
[final product], [information output], [intermediary], [department], [work
group], [customer],
[co-customer], [customer of customer], [procurement], [transportation],
[storage], [design],
[production], [quality control], [sales], [exchange], [banking], [investment
design],
[management [audit], [capital], [energy], [information], [land], [tools], or
[labor].
[245] In addition, proximity relationships may be derived from the non-
proprietary,
commonly available tags, attributes, and values assigned to the economic
entities. As non-
limiting examples, these tags, attributes, and values may include [asset
class], [exchange
listing], [yield], [duration], [convexity], [date founded], [location of
headquarters], [location
of incorporation], [market capitalization], [revenue], [expenses], [net
income], [cash flow
from operations], [cash flow from financing], or [cash flow from investing].
[246] These tags, attributes, and values may be weighted, score, timestamped,
or geotagged,
and stored in one or more databases as a basis for proximity calculations. In
some
embodiments, economic entities or securities currently sharing numerous
identical or similar
tags, attributes, and values will be proximate within one or more databases
used to provide
recommendations, while economic entities with few common tags will be distant
within those
databases.
[247] The relationships, tags, attributes, and/or values, derived from one or
more databases
of securities and economic entities, permit a default calculation of proximity
to a user. User
53
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
preferences, current user holdings, and network position facilitate the
customization of
financial recommendations to users based on dynamic proximity calculations.
[248] In some embodiments, users may input their express preferences, whether
syntactic,
functional, non-syntactic, non-functional, or some combination thereof, and
associated values
into the system upon registering to gain access to the database. In other
embodiments, these
preferences, or filters, may be inputted or modified at any time, either
through a separate
module or by indicating a preference for or against data entities associated
with securities and
economic entities. The filters may be absolute, in that they will permit the
user to exclude or
include certain relationships, attributes, tags, or values, or they may be
relative, in that they
enable the user to indicate the extent of a preference for or against certain
relationships,
attributes, tags, or values.
[249] As non-limiting examples, these filters may enable the user to express
absolute or
relative preferences for [market capitalization], [asset class], [asset
allocation], [funds],
[expected return], [risk], [geography], [supplier], [investor], [customer],
[lender-borrower],
[issuer-investor], [1.1], [1.2], [1.3], [2.1], [2.2], [2.3], [3.1], [3.2],
[3.3], [4.1], [4.2], [4.3],
[1], [.2], [.3], [Al, [B], [C], [D], [E], [F], [li], [lii], [liii], [2i],
[2ii], [2iii], [3i], [3ii], [3iii],
[4i], [4ii], [4iii], [portfolio], [composite], [stratified structure], or one
or more of any of the
other relationships, tags, attributes, or values assigned to the securities
and economic entities.
[250] Users may also reveal their preferences through their interactions with
the data
entities on the system. In some embodiments, preferences will be revealed by
tracking user
accounts, monitoring clicks, screen time, portfolio construction, and/or
transactions executed,
and using a machine learning process to improve dynamically the customized
recommendations to a user based on their preferences. In some embodiments,
users may
upload their portfolios to the system, whose constituents also may be used to
guide
customized recommendations.
[251] Network position may facilitate proximity calculations and dynamic,
customized
recommendations. The system may track connections among users and the extent
of their
interactions. In some embodiments, strong connections among users on the
system will lead
their recommendations to converge significantly, weak connections will lead
the
recommendations to converge slightly, and numerous degrees of separation will
lead the
recommendations to diverge.
54
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[252] In some embodiments, similarities among users in the system may lead the
recommendations provided to them to converge, while differences among the
users may lead
those recommendations to diverge. As a non-limiting example, the system may
use machine
learning techniques to improve dynamically the quality of customized
recommendations
based on changes in the network of users, their preferences, or the tags,
attributes, values, or
relationships assigned to the securities or economic entities.
[253] Data Analytics
[254] The systems syntax described herein is well-suited to organize and
analyze very large
data sets associated with domains that can be effectively studied through
functional models,
including, as non-limiting examples, biology, physics, ecology, economics,
computer science,
genomics, bioinformatics, aeronautics, telecommunications, electrodynamics,
astronautics,
finance, investment management, healthcare, medicine, epidemiology, chemistry,
geology,
transportation, engineering, legal systems, regulatory systems, legislative
systems, political
systems, and economic development. As non-limiting examples, data sets
characterizing
these complex systems may be hundreds of terabytes or petabytes in size, have
hundreds of
thousands or millions of elements, and have thousands of variables that
significantly impact
the characteristics or features of the system. The analysis of these complex
systems is
impracticable without an underlying functional model and advanced customized
data
systems.
[255] The assignment of tags, metatags, attributes, and values derived from an
underlying
relational model of activities and resources in complex systems facilitates
the development of
real-time tools to enable diagnostics, customized recommendations, and
predictive analytics,
thereby permitting dynamic responses to rapidly changing events. These real-
time tools may
be particularly critical during periods in which the system is chaotic or far
from equilibrium;
as non-limiting examples, these may include perturbations, shocks, natural
disasters, bubbles,
panics, manias, or crashes, periods during which mechanical models of systems
and standard
database tools are likely to prove ineffective or even harmful. As non-
limiting examples, the
tags, metatags, or attributes may be syntactic, semantic, morphosyntatic,
morphological,
physiological, anatomical, geographic, temporal, or demographic. In some
embodiments, the
assignment or identification of one or more tags, attributes, or values, or
the proximity or
similarity of tags, attributes, or values, may facilitate the assignment,
identification,
prediction, or recommendation of one or more other tags, attributes, or
values. In other
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
embodiments, the application of graph structures or network models facilitates
the
development of these analytical tools.
[256] Normative Cases for Stratified Portfolios
[257] Using the systems and methods described herein, a normative stratified
or segmented
portfolio can be defined. Stratified or segmented units can be used as a tool
for building
normative models and developing normative target scores. Reliable and
validated categories
of investment securities can be used to sub-divide populations of securities
to validate
normative studies. The user can develop normative scores to test a hypothesis
and validate a
baseline for use in the comparative study of other stratified or segmented
portfolios. The
system can be configured so that a normative stratified portfolio can be used
to derive a target
score. A target score for a stratified portfolio, such as a target alpha
score, can be defined
relative to a baseline normative target score.
[258] A variety of statistical properties may be studied using empirical or
simulated data on
a stratified or segmented portfolio, group, or subgroup. As non-limiting
examples, a statistical
property may be selected from among mean, variance, standard deviation, skew,
kurtosis,
correlation, semivariance, and semideviation, or the excess or residual of any
of these.
[259] Statistical tests can be used to establish that for securities
associated with attribute-
defined functional groupings of companies, commodities, securities, funds,
assets, loans, or
liabilities a) they exhibit higher intra-group correlation than inter-group
correlation; b) that
correlation is more persistent and predictive over time than covariance in
groupings created
quantitatively; c) those groupings can be segmented or stratified in a
portfolio to target or
control for particular exposures to volatility, variance, or non-systematic
risk; d) the variance,
standard deviation, semideviation, and/or semivariance are lower at the
portfolio level than at
the group level, and lower at the group level than at the security level; e)
for a given
performance metric, exhibit more normally distributed expected or actual
values than an
alternative grouping, index, or portfolio; or 0 this methodology increases the
predictability of
outcomes and the extent to which returns on large portfolios consistently
achieve those
predicted by theory.
[260] In some embodiments, the performance metrics may include performance,
volatility,
liquidity, variance, expected return, alpha, Jensen's alpha, beta, variance,
covariance,
semivariance, semideviation, correlation, autocorrelation, Sharpe ratio,
Sortino ratio, revenue,
expenses, operating expenses, earnings, net earnings, gross earnings, income,
gross income,
56
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
net income, cash flow, cash flow from operations, cash flow from operations,
cash flow from
investing, or cash flow from financing. As non-limiting examples, the
normality may be
assessed using the Cramer¨von Mises criterion, the Kolmogorov-Smirnov test,
the Shapiro-
Wilk test, the Anderson-Darling test, the Jarque-Bera test, the Siegel-Tukey
test, Kuiper test,
a p-value test, a Q-Q plot, a test of skewness, or a test of kurtosis.
[261] As non-limiting examples, this statistical methodology enables the
construction of
large and mid-cap equity portfolios that achieve a consistent risk premium to
debt over
extended periods of time, as predicted by the Capital Asset Pricing Model, and
permits the
development of indices that realize rates of risk and liquidity-adjusted
return that more
predictably attain the market performance posited by financial theory than
indices such as the
S&P 500TM which are frequently used as proxies for the market.
[262] At an initial step, one or more theoretical or estimated scores can be
defined. Using
adjustments based on changes made to at least one of the following: 1) changes
to the
population of investment securities; 2) the stratification or segmentation
methodology applied
to the population of investment securities; and 3) the weighting applied to
the stratified units
or segments, the portfolio can be engineered to: 1) create a representative
outcome for a given
population (referred to herein as a normative case); 2) engineer an outcome
that is statistically
biased in a user-specified direction.
[263] Depending on the adjustment methodology, the bias can be towards a
population
subset such as a geographic or temporal group or a particular functional
attribute class (or
subset of an attribute class) within a specific population set of securities.
Within a stratified
or segmented architecture for a given population, a specific exposure (or lack
thereof) can be
managed through the structure itself (either through structure or attribute
selection) or the
weighting assigned to specific units, groups, attributes, clusters, or
segments.
[264] Non-normative composites are composites that are designed to vary from
the
normative case. Divergence from the normative case may be considered to be an
engineered
or algorithmic portfolio performance metric, e.g. alpha. Using the invention,
negative
variance can be engineered as alpha for short investment positions.
Engineering positive
variance can be engineered as alpha for long investment positions. For
example, distributions
can be normal (based on the normative case) or non-normal. Non-normative
distributions can
be positively skewed (to the right of normal), negatively skewed (to the left
of normal),
57
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
platykurtic (fat-tailed), or leptokurtic (thin-tailed). Adjustments to the
weightings, as
described above, can be used to generate portfolios having these types of
distributions.
[265] In some embodiments, groupings can be used to establish pairwise
correlation
coefficients to be used in Markowitz mean-variance optimization. Instead of
using a single
correlation value for all pairs of companies, the method can be used to assign
correlations to
any segment created through this methodology, by, as a non-limiting example,
taking the
average observed pairwise residual correlation of all segments in a stratum,
as well as a
measure of out-of-group correlations.
[266] As a non-limiting example, estimating a correlation value for each
segment in the
third stratum of a stratified heterogeneous 900 security portfolio or index
and assigning the
relevant value to each constituent of a group can reduce the number of
correlations necessary
to estimate pairwise correlations by a factor of over 200, facilitating the
construction of an
index or portfolio that will more consistently and predictably approximate the
efficient
frontier than other methods of portfolio or index construction.
[267] Data Set Normalization and Probability Shaping
[268] Mathematical processing according to the methods described herein can be
applied to
large sets of economic and financial data to reduce these fluctuations and
randomness of the
results, including, as a non-limiting example, those of investment returns. In
some
embodiments, they include multivariate algorithms can be used to organize
large datasets.
The methods can be used to generate or identify causal connections and perform
real-time
analyses.
[269] The system can be configured for normalizing the data sets representing
securities.
The normalization process includes statistical categorization based on
attributes of the entity
associated with the security. The attributes used for normalization can be
those types of
attributes described above, or other attributes relating, as non-limiting
examples, to the
operations, assets, suppliers, customers, customers of customers, departments,
or employees
of the entity associated with the security.
[270] Multiple investment securities can be organized into statistical
categories. A user
interface for selecting among the attributes can be provided by the system,
which can include
a statistical categories editor (referred to as a thesaurus editor in some
embodiments). The
statistical categories can be defined within the system using the editor. A
statistical category
can be defined to be any one or more of the attributes described above, taken
alone or in
58
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
combination with one other. The statistical categories also can be defined
based on the
syntax and coding systems described above. In some cases, a statistical
category can also be
a stratified or segmented unit.
[271] Portfolio Powering
[272] Matching or stratification also improves statistical power, particularly
if matching,
segmentation, or stratification is based on important prognostic variables.
Such procedures,
accompanied by pre-specified stratified or segmented analyses and sensitivity
analyses may,
therefore, be useful.
[273] A prospective analysis can be used to determine a sample size required
to achieve
target statistical power. In general, the most important component affecting
statistical power
is sample size in the sense that the most frequently asked question in
practice is how many
observations need to be collected. As a non-limiting example, in assessing
portfolio
performance, the null hypothesis could be that a stratified grouping has a
Sharpe ratio of 1.
The alternative hypothesis could be that a stratified grouping has Sharpe
ratio other than 1.
[274] Power refers to the probability that a test will find a statistically
significant difference
when such a difference actually exists. Power is the probability of correctly
rejecting the null
hypothesis. In some embodiments, power should be .8 or greater such that there
is an 80% or
greater chance of finding a statistically significant difference when there is
one.
[275] Bankruptcy Example
[276] The following example illustrates a use case for a composite of
investment securities.
In this example, a stratified or segmented composite portfolio of investment
grade corporate
debt securities is created.
[277] Investment-grade debt is a specific class of securities with a well-
defined expected
rate of return and a well-defined risk. Each bond is rated by a third-party
rating agency. This
rating captures the estimated likelihood that the bond issuer will default on
the debt. In the
case of default risk, one of the most pertinent risks in investing in such
securities, corporate
bonds with the same rating should have similar yields to maturity, holding
other variables,
such as maturity, constant. The yield to maturity is the compounded annual
rate of return that
the bondholder will earn in holding the bond to maturity given its current
price, assuming all
payments (coupon payments and face value) are made as expected. Put another
way, the
yield to maturity is the discount rate that makes the present value of the
bond's future cash
59
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
flows, assuming all payments are made, equal to the current price of the bond.
For all bonds
that have a comparable rating from these agencies, the forecasted yields for a
given maturity
date will be the same or within a very tight range. That is, investment-grade
corporate debt
securities behave predictably.
[278] While different investment-grade debt securities may have the same
estimated
probability of default, the event or events that trigger a default vary from
issuer to issuer.
That is, companies may face different risk factors relative to the specific
attribute set
associated with the company and its operations. Some of these factors may be
unique to that
company, while others may be common to groups of companies. Such risks may
include, as
non-limiting examples, industry risk, product risk, customer risk, sensitivity
to interest rates,
geographic, political, or economic factors outside a company's control, or
risks related to the
company's CEO or management in general. There are many company-specific
attributes that
can be tied to a company's default risk. These can include, but are not
limited to:
[279] 1) Functional operating or asset-based attributes: Such attributes are
not accounting or
performance measures and indicators, but rather, as non-limiting examples,
attributes that
define what a company does, such as manufacturing or transportation;
attributes or tags
related to the company's product, such as car, computer, or couch as well as
type of car,
computer or couch; attributes related to a company's customer such as consumer
or business;
attributes related to the customer's customer; attributes related to the
geographic location of a
business or its individual operations; attributes related to the products and
materials a
company uses to provide its product; attributes related to any of the
multivariate industries or
industry segments in which a company may operate; attributes related to the
structure of a
company's business such as integrated, non-integrated, forward integrated,
backward
integrated, or networked; attributes related to any of the multivariate
governmental or
macroeconomic risks associated with a specific business or country where a
company does
business; attributes associated with the accounting or business risks listed
by a company as
core to their business; risks associated with categorization tied to a
specific business or
segment by the investment community. At any given point in time, any one of
these factors or
industry events related to these factors may cause or increase the risk of
bankruptcy in any
specific company.
[280] 2) Management or strategy: A company has unique risks based on its
management
team, its decisions, and its strategies.
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[281] 3) Company asset value: Bankruptcy (being one type of default)
fundamentally
changes the terms of the securities issued by a single company. Upon filing,
the presumption
of returns based on ongoing operations changes to include a liquidation
scenario and the
analysis of the rights of each security holder. In this case, investors assess
their ability to
receive payment on a given security based on its location in the corporate
capital structure.
Securities may have been assigned a priority in liquidation. If an underlying
asset of a
company is sold or disposed of, these liquidation priorities designate
seniority.
[282] 4) Financial leverage: Some companies are more or less levered than
other
companies.
[283] Each attribute is a potential source of default or bankruptcy risk for a
fixed-income
investor. Some of these attributes may relate to groups of companies (e.g.
companies that
produce cars, or companies whose operations are located in New Orleans).
Because of this, a
portfolio that does not control for specific attributes can be inadvertently
exposed to a
concentration in a specific risk. When a member of a group defaults or files
for bankruptcy,
other companies in that group may also be impacted.
[284] The invention includes methods for building a stratified or segmented
composite
portfolio of investment-grade corporate debt in such a way that limits
exposure to bankruptcy
risks, corporate events, and other such non-systematic risk factors by
managing the
portfolio's exposure to any particular company or industry. In capitalization-
weighted debt
portfolios, securities are weighted in proportion to their issuance size
relative to the total size
of all issues in the portfolio. With such an unmanaged weighting scheme, it is
possible for
companies or industries that issue large amounts of debt to become over-
weighted in the
portfolio. If one of these companies or industries has a negative event, such
as bankruptcy,
then the portfolio itself will be dramatically impacted. A stratified or
segmented composite
portfolio is a tool to cap financial exposure to attribute-related risks.
[285] The application of the invention to manage default risk in an investment-
grade
corporate debt portfolio provides an illustration of one embodiment. Each debt
security has a
level of risk that is directly tied to the value in liquidation of the
underlying assets of the
company. This risk is distinctly separate from financial market risks
associated with the
supply and demand of the debt security itself, as well as from financial
market factors that
may impact the rate of return needed for a given investment security at a
given point in time,
such as the risk free rate at that point in time.
61
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[286] The systems described herein protect against such non-systematic risks
across the
portfolio; that is, they can reduce or eliminate material impacts of a single
security or group
of securities. This can be achieved by organizing companies in groups (strata
or segments)
based on non-systematic attributes, e.g. by grouping together companies with
similar
products, or similar customer bases. In some cases, stratification or
segmentation ensures
that no single non-systematic exposure represents a material risk to the
portfolio as a whole.
In such a composite, bankruptcy exposure is spread across enough unique groups
to minimize
the impact of bankruptcies in any one group or company.
[287] As a non-limiting example, the invention can be used to create strata or
segments as
follows. For investment grade bonds, there may be several types of causes of a
downgrade or
bankruptcy, which may include, as non-limiting examples: 1) company-specific
exposure; 2)
industry-specific exposure; and 3) product-specific exposure. Investment-grade
bonds of a
given rating theoretically should have the same probability ex ante of
downgrade or
bankruptcy risk, but this rating provides no information about the probable
causes for
bankruptcy. And indeed, for bonds of the same rating, the factors that may
cause the issuer to
default can be radically different. These bankruptcy factors, however, are
directly linked to
the functional attributes of the issuing company. Using these attributes, it
is possible to group
bonds into risk groups based on the properties of their issuers that relate to
issuers'
bankruptcy factors. This process may be repeated to form a nested architecture
of groups,
where each sub-group has its own risk but also has risks associated with the
parent group. It
also may be repeated to form graphs or networks of segmented groups, where
each segment
and sub-segment share common risks. These risk groups, then, are the strata or
segments that
may be used to construct a stratified or segmented investment composite,
respectively. These
processes reduce or mitigate the chance that a negative event in either a
single company or
industry can severely impact the portfolio.
[288] Industry Risk Example
[289] The following example illustrates an additional use case for stratified
or segmented
composites of investment securities. In this example, a composite of equities
from the S&P
900 index is created. This composite is a broad-based index comprising large-
and mid-cap
equities issued by US-headquartered companies from a variety of industries.
This universe is
a combination of the S&P 500 and S&P MidCap 400 indexes, which track large-
and
mid-cap US companies, respectively. Over periods of time, such a universe of
equities
62
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
should display a consistent return premium relative to a relatively risk-free
investment such
as US Treasury Bills.
[290] In this example, the returns of the capitalization-weighted S&P 900 are
compared
with the returns of the same universe of securities engineered into a
stratified composite
constructed using the method of the invention. Attributes relating to the
functional
characteristics of these 900 companies are used to create nested strata that
group functionally
similar companies together. These strata are used to determine weights for
each security
following the methods described herein. The portfolio is rebalanced quarterly,
returning each
security to its initial weight.
[291] Stratification and segmentation provide material benefits in
environments when
specific industries experience large negative price shocks, colloquially
referred to as an
industry bubble "bursting". As an industry bubble grows, the market
capitalization of the
companies in the industry grows, thus increasing that industry's weight in the
capitalization-
weighted portfolio. In capitalization-weighted funds, which lack attribute-
based controls on
the weights of both individual companies and groups of similar companies, such
bubbles can
create unintended overexposure to specific risk groups, including those that
disproportionately impact a particular industry. When the over-weighted
industry bubble
collapses, the portfolio suffers disproportionately. Even if the companies
outside of the
industry bubble perform reasonably, the negative returns of the over-weighted
companies can
result in negative returns for the entire portfolio.
[292] In stratified composite portfolios, however, the risk of industry
bubbles can be
substantially mitigated by stratifying the universe such that the strata
correspond to distinct
industry risks. In this manner, industry-specific risks are isolated and
cannot induce
disproportionately negative performance in the portfolio.
[293] The growth and collapse of information technology equities from 1997 to
2000
exemplifies the benefits of stratified composite portfolios. Using functional
attributes, a
group of companies whose business function involves moving, storing, or
processing
information is defined. Companies in this group include Microsoft, Cisco,
Intel, AOL,
Qualcomm, and other such information technology companies.
[294] The twenty largest such information technology equities in the S&P 900
grew in
weight over the late 1990s such that by the year 2000, they dominated the
portfolio. At
yearend 1997, 1998, and 1999, these twenty equities collectively weighed
11.8%, 13.7%, and
63
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
20.4% of the S&P 900t, respectively. In 2000, when the bubble collapsed, these
equities fell
in value by 42.3%, while the S&P 900 as a whole returned -6.9%. Excluding
these
information companies, the rest of the S&P 900 returned 6.8%. That is, the
"market-wide"
downturn in 2000 was not a systematic failure; it was the result of
uncontrolled over-
exposure to a single industry.
[295] In a stratified composite portfolio, such industry-specific risk can be
controlled. In
the example stratified composite portfolio, the same twenty information
companies were set
at a weight of 2.9% and were rebalanced to this weight quarterly. In 2000,
this isolated group
performed poorly (falling in value by 59.7%), but outside of this group, the
example stratified
composite portfolio had healthy returns. Excluding these twenty companies, the
example
stratified composite portfolio returned 21.3%. In total, the example
stratified composite
portfolio returned 17.6% in the year 2000, outperforming the capitalization-
weighted
portfolio of the exact same universe by 24.5%.
[296] The performance of the capitalization-weighted S&P 900 against the
example
stratified composite portfolio of the same universe demonstrates how
stratification can
prevent non-systematic industry risks from impacting an entire portfolio.
[297] System Architectures
[298] The systems and methods described herein can be implemented in software
or
hardware or any combination thereof The systems and methods described herein
can be
implemented using one or more computing devices which may or may not be
physically or
logically separate from each other. The methods may be performed by components
arranged
as either on-premise hardware, on-premise virtual systems, or hosted-private
instances.
Additionally, various aspects of the methods described herein may be combined
or merged
into other functions.
[299] An example logical implementation of the system is illustrated in Fig.
10.
Relationships between tables (1005, 1010, 1015, 1020, 1025, and 1030) are
illustrated by
arrows. As illustrated, table 1010 serves as a linking table between companies
table (1005)
and barcode table (1015).
[300] An example computerized system for implementing the invention is
illustrated in Fig.
11. A processor or computer system can be configured to particularly perform
some or all of
the method described herein. In some embodiments, the method can be partially
or fully
automated by one or more computers or processors. The invention may be
implemented
64
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
using a combination of any of hardware, firmware and/or software. The present
invention (or
any part(s) or function(s) thereof) may be implemented using hardware,
software, firmware,
or a combination thereof and may be implemented in one or more computer
systems or other
processing systems. In some embodiments, the illustrated system elements could
be
combined into a single hardware device or separated into multiple hardware
devices. If
multiple hardware devices are used, the hardware devices could be physically
located
proximate to or remotely from each other. The embodiments of the methods
described and
illustrated are intended to be illustrative and not to be limiting. For
example, some or all of
the steps of the methods can be combined, rearranged, and/or omitted in
different
embodiments.
[301] In one exemplary embodiment, the invention may be directed toward one or
more
computer systems capable of carrying out the functionality described herein.
Example
computing devices may be, but are not limited to, a personal computer (PC)
system running
any operating system such as, but not limited to, MicrosoftTM WindowsTM.
However, the
invention may not be limited to these platforms. Instead, the invention may be
implemented
on any appropriate computer system running any appropriate operating system.
Other
components of the invention, such as, but not limited to, a computing device,
a
communications device, mobile phone, a telephony device, a telephone, a
personal digital
assistant (PDA), a personal computer (PC), a handheld PC, an interactive
television (iTV), a
digital video recorder (DVD), client workstations, thin clients, thick
clients, proxy servers,
network communication servers, remote access devices, client computers, server
computers,
routers, web servers, data, media, audio, video, telephony or streaming
technology servers,
etc., may also be implemented using a computing device. Services may be
provided on
demand using, e.g., but not limited to, an interactive television (iTV), a
video on demand
system (VOD), and via a digital video recorder (DVR), or other on demand
viewing system.
[302] The system may include one or more processors. The processor(s) may be
connected
to a communication infrastructure, such as but not limited to, a
communications bus, cross-
over bar, or network, etc. The processes and processors need not be located at
the same
physical locations. In other words, processes can be executed at one or more
geographically
distant processors, over for example, a LAN or WAN connection. Computing
devices may
include a display interface that may forward graphics, text, and other data
from the
communication infrastructure for display on a display unit.
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[303] The computer system may also include, but is not limited to, a main
memory, random
access memory (RAM), and a secondary memory, etc. The secondary memory may
include,
for example, a hard disk drive and/or a removable storage drive, such as a
compact disk drive
CD-ROM, etc. The removable storage drive may read from and/or write to a
removable
storage unit. As may be appreciated, the removable storage unit may include a
computer
usable storage medium having stored therein computer software and/or data. In
some
embodiments, a machine-accessible medium may refer to any storage device used
for storing
data accessible by a computer. Examples of a machine-accessible medium may
include, e.g.,
but not limited to: a magnetic hard disk; a floppy disk; an optical disk, like
a compact disk
read-only memory (CD-ROM) or a digital versatile disk (DVD); a magnetic tape;
and/or a
memory chip, etc.
[304] The processor may also include, or be operatively coupled to communicate
with, one
or more data storage devices for storing data. Such data storage devices can
include, as non-
limiting examples, magnetic disks (including internal hard disks and removable
disks),
magneto-optical disks, optical disks, read-only memory, random access memory,
and/or flash
storage. Storage devices suitable for tangibly embodying computer program
instructions and
data can also include all forms of non-volatile memory, including, for
example,
semiconductor memory devices, such as EPROM, EEPROM, and flash memory devices;
magnetic disks such as internal hard disks and removable disks; magneto-
optical disks; and
CD-ROM and DVD-ROM disks. The processor and the memory can be supplemented by,
or
incorporated in, ASICs (application-specific integrated circuits).
[305] The processing system can be in communication with a computerized data
storage
system. The data storage system can include a non-relational or relational
data store, such as
a MySQLTM or other relational database. Other physical and logical database
types could be
used. The data store may be a database server, such as Microsoft SQL ServerTM,
OracleTM,
IBM DB2TM, SQLITETm, or any other database software, relational or otherwise.
The data
store may store the information identifying syntactical tags and any
information required to
operate on syntactical tags. In some embodiments, the processing system may
use object-
oriented programming and may store data in objects. In these embodiments, the
processing
system may use an object-relational mapper (ORM) to store the data objects in
a relational
database. The systems and methods described herein can be implemented using
any number
of physical data models. In one example embodiment, an RDBMS can be used. In
those
embodiments, tables in the RDBMS can include columns that represent
coordinates. In the
66
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
case of economic systems, data representing companies, products, etc. can be
stored in tables
in the RDBMS. The tables can have pre-defined relationships between them. The
tables can
also have adjuncts associated with the coordinates.
[306] In alternative exemplary embodiments, secondary memory may include other
similar
devices for allowing computer programs or other instructions to be loaded into
computer
system. Such devices may include, for example, a removable storage unit and an
interface.
Examples of such may include a program cartridge and cartridge interface (such
as, e.g., but
not limited to, those found in video game devices), a removable memory chip
(such as, e.g.,
but not limited to, an erasable programmable read only memory (EPROM), or
programmable
read only memory (PROM) and associated socket, and other removable storage
units and
interfaces, which may allow software and data to be transferred from the
removable storage
unit to computer system.
[307] The computing device may also include an input device such as but not
limited to, a
mouse or other pointing device such as a digitizer, and a keyboard or other
data entry device
(not shown). The computing device may also include output devices, such as but
not limited
to, a display, and a display interface. Computer may include input/output
(I/O) devices such
as but not limited to a communications interface, cable and communications
path, etc. These
devices may include, but are not limited to, a network interface card, and
modems.
Communications interface may allow software and data to be transferred between
computer
system and external devices.
[308] In one or more embodiments, the present embodiments are practiced in the
environment of a computer network or networks. The network can include a
private network,
or a public network (for example the Internet, as described below), or a
combination of both.
The network includes hardware, software, or a combination of both.
[309] From a telecommunications-oriented view, the network can be described as
a set of
hardware nodes interconnected by a communications facility, with one or more
processes
(hardware, software, or a combination thereof) functioning at each such node.
The processes
can inter-communicate and exchange information with one another via
communication
pathways between them using interprocess communication pathways. On these
pathways,
appropriate communications protocols are used.
[310] An exemplary computer and/or telecommunications network environment in
accordance with the present embodiments may include node, which include may
hardware,
67
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
software, or a combination of hardware and software. The nodes may be
interconnected via a
communications network. Each node may include one or more processes,
executable by
processors incorporated into the nodes. A single process may be run by
multiple processors,
or multiple processes may be run by a single processor, for example.
Additionally, each of
the nodes may provide an interface point between network and the outside
world, and may
incorporate a collection of sub-networks.
[311] In an exemplary embodiment, the processes may communicate with one
another
through interprocess communication pathways supporting communication through
any
communications protocol. The pathways may function in sequence or in parallel,
continuously or intermittently. The pathways can use any of the communications
standards,
protocols or technologies, described herein with respect to a communications
network, in
addition to standard parallel instruction sets used by many computers.
[312] The nodes may include any entities capable of performing processing
functions.
Examples of such nodes that can be used with the embodiments include computers
(such as
personal computers, workstations, servers, or mainframes), handheld wireless
devices and
wireline devices (such as personal digital assistants (PDAs), modem cell
phones with
processing capability, wireless email devices including BlackBerryTM devices),
document
processing devices (such as scanners, printers, facsimile machines, or
multifunction
document machines), or complex entities (such as local-area networks or wide
area networks)
to which are connected a collection of processors, as described. For example,
in the context
of the present invention, a node itself can be a wide-area network (WAN), a
local-area
network (LAN), a private network (such as a Virtual Private Network (VPN)), or
collection
of networks.
[313] Communications between the nodes may be made possible by a
communications
network. A node may be connected either continuously or intermittently with
communications network. As an example, in the context of the present
invention, a
communications network can be a digital communications infrastructure
providing adequate
bandwidth and information security.
[314] The communications network can include wireline communications
capability,
wireless communications capability, or a combination of both, at any
frequencies, using any
type of standard, protocol or technology. In addition, in the present
embodiments, the
68
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
communications network can be a private network (for example, a VPN) or a
public network
(for example, the Internet).
[315] A non-inclusive list of exemplary wireless protocols and technologies
used by a
communications network may include BlueToothTm, general packet radio service
(GPRS),
cellular digital packet data (CDPD), mobile solutions platform (MSP),
multimedia messaging
(MMS), wireless application protocol (WAP), code division multiple access
(CDMA), short
message service (SMS), wireless markup language (WML), handheld device markup
language (HDML), binary runtime environment for wireless (BREW), radio access
network
(RAN), and packet switched core networks (PS-CN). Also included are various
generation
wireless technologies. An exemplary non-inclusive list of primarily wireline
protocols and
technologies used by a communications network includes asynchronous transfer
mode
(ATM), enhanced interior gateway routing protocol (EIGRP), frame relay (FR),
high-level
data link control (HDLC), Internet control message protocol (ICMP), interior
gateway
routing protocol (IGRP), internetwork packet exchange (IPX), ISDN, point-to-
point protocol
(PPP), transmission control protocol/internet protocol (TCP/IP), routing
information protocol
(RIP) and user datagram protocol (UDP). As skilled persons will recognize, any
other known
or anticipated wireless or wireline protocols and technologies can be used.
[316] Embodiments of the present invention may include apparatuses for
performing the
operations herein. An apparatus may be specially constructed for the desired
purposes, or it
may comprise a general purpose device selectively activated or reconfigured by
a program
stored in the device.
[317] In one or more embodiments, the present embodiments are embodied in
machine-
executable instructions. The instructions can be used to cause a processing
device, for
example a general-purpose or special-purpose processor, which is programmed
with the
instructions, to perform the steps of the present invention. Alternatively,
the steps of the
present invention can be performed by specific hardware components that
contain hardwired
logic for performing the steps, or by any combination of programmed computer
components
and custom hardware components. For example, the present invention can be
provided as a
computer program product, as outlined above. In this environment, the
embodiments can
include a machine-readable medium having instructions stored on it. The
instructions can be
used to program any processor or processors (or other electronic devices) to
perform a
process or method according to the present exemplary embodiments. In addition,
the present
69
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
invention can also be downloaded and stored on a computer program product.
Here, the
program can be transferred from a remote computer (e.g., a server) to a
requesting computer
(e.g., a client) by way of data signals embodied in a carrier wave or other
propagation
medium via a communication link (e.g., a modem or network connection) and
ultimately such
signals may be stored on the computer systems for subsequent execution).
[318] The methods can be implemented in a computer program product accessible
from a
computer-usable or computer-readable storage medium that provides program code
for use
by or in connection with a computer or any instruction execution system. A
computer-usable
or computer-readable storage medium can be any apparatus that can contain or
store the
program for use by or in connection with the computer or instruction execution
system,
apparatus, or device.
[319] A data processing system suitable for storing and/or executing the
corresponding
program code can include at least one processor coupled directly or indirectly
to
computerized data storage devices such as memory elements. Input/output (I/O)
devices
(including but not limited to keyboards, displays, pointing devices, etc.) can
be coupled to the
system. Network adapters may also be coupled to the system to enable the data
processing
system to become coupled to other data processing systems or remote printers
or storage
devices through intervening private or public networks. To provide for
interaction with a
user, the features can be implemented on a computer with a display device,
such as an LCD
(liquid crystal display), or another type of monitor for displaying
information to the user, and
a keyboard and an input device, such as a mouse or trackball by which the user
can provide
input to the computer.
[320] A computer program can be a set of instructions that can be used,
directly or
indirectly, in a computer. The systems and methods described herein can be
implemented
using programming languages such as FlashTM, JAVATM, C++, C, C#, Python,
Visual
BasicTM, JavaScriptTM PHP, XML, HTML, etc., or a combination of programming
languages,
including compiled or interpreted languages, and can be deployed in any form,
including as a
stand-alone program or as a module, component, subroutine, or other unit
suitable for use in a
computing environment. The software can include, but is not limited to,
firmware, resident
software, microcode, etc. Protocols such as SOAP/HTTP may be used in
implementing
interfaces between programming modules. The components and functionality
described
herein may be implemented on any desktop operating system executing in a
virtualized or
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
non-virtualized environment, using any programming language suitable for
software
development, including, but not limited to, different versions of Microsoft
WindowsTM,
AppleTM MacTM, iOSTM, UnixTm/X-WindowsTm, LinuxTM, etc. The system could be
implemented using a web application framework, such as Ruby on Rails.
[321] Suitable processors for the execution of a program of instructions
include, but are not
limited to, general and special purpose microprocessors, and the sole
processor or one of
multiple processors or cores, of any kind of computer. A processor may receive
and store
instructions and data from a computerized data storage device such as a read-
only memory, a
random access memory, both, or any combination of the data storage devices
described
herein. A processor may include any processing circuitry or control circuitry
operative to
control the operations and performance of an electronic device.
[322] The systems, modules, and methods described herein can be implemented
using any
combination of software or hardware elements. The systems, modules, and
methods
described herein can be implemented using one or more virtual machines
operating alone or
in combination with one other. Any applicable virtualization solution can be
used for
encapsulating a physical computing machine platform into a virtual machine
that is executed
under the control of virtualization software running on a hardware computing
platform or
host. The virtual machine can have both virtual system hardware and guest
operating system
software.
[323] The systems and methods described herein can be implemented in a
computer system
that includes a back-end component, such as a data server, or that includes a
middleware
component, such as an application server or an Internet server, or that
includes a front-end
component, such as a client computer having a graphical user interface or an
Internet
browser, or any combination of them. The components of the system can be
connected by
any form or medium of digital data communication such as a communication
network.
Examples of communication networks include, e.g., a LAN, a WAN, and the
computers and
networks that form the Internet.
[324] One or more embodiments of the invention may be practiced with other
computer
system configurations, including hand-held devices, microprocessor systems,
microprocessor-based or programmable consumer electronics, minicomputers,
mainframe
computers, etc. The invention may also be practiced in distributed computing
environments
where tasks are performed by remote processing devices that are linked through
a network.
71
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
[325] The terms "computer program medium" and "computer readable medium" may
be
used to generally refer to media such as but not limited to removable storage
drive, a hard
disk installed in hard disk drive. These computer program products may provide
software to
computer system. The invention may be directed to such computer program
products.
[326] References to "one embodiment," "an embodiment," "example embodiment,"
"various embodiments," etc., may indicate that the embodiment(s) of the
invention so
described may include a particular feature, structure, or characteristic, but
not every
embodiment necessarily includes the particular feature, structure, or
characteristic. Further,
repeated use of the phrase "in one embodiment," or "in an exemplary
embodiment," do not
necessarily refer to the same embodiment, although they may.
[327] In the description and claims, the terms "coupled" and "connected,"
along with their
derivatives, may be used. It should be understood that these terms may be not
intended as
synonyms for each other. Rather, in particular embodiments, "connected" may be
used to
indicate that two or more elements are in direct physical or electrical
contact with each other.
"Coupled" may mean that two or more elements are in direct physical or
electrical contact.
However, "coupled" may also mean that two or more elements are not in direct
contact with
each other, but yet still co-operate or interact with each other.
[328] An algorithm may be here, and generally, considered to be a self-
consistent sequence
of acts or operations leading to a desired result. These include physical
manipulations of
physical quantities. Usually, though not necessarily, these quantities take
the form of
electrical or magnetic signals capable of being stored, transferred, combined,
compared, and
otherwise manipulated. It has proven convenient at times, principally for
reasons of common
usage, to refer to these signals as bits, values, elements, symbols,
characters, terms, numbers
or the like. It should be understood, however, that all of these and similar
terms are to be
associated with the appropriate physical quantities and are merely convenient
labels applied
to these quantities.
[329] Unless specifically stated otherwise, it may be appreciated that
throughout the
specification terms such as "processing," "computing," "calculating,"
"determining," or the
like, refer to the action and/or processes of a computer or computing system,
or similar
electronic computing device, that manipulate and/or transform data represented
as physical,
such as electronic, quantities within the computing system's registers and/or
memories into
72
CA 02974701 2017-07-21
WO 2016/118939
PCT/US2016/014642
other data similarly represented as physical quantities within the computing
system's
memories, registers or other such information storage, transmission or display
devices.
[330] In a similar manner, the term "processor" may refer to any device or
portion of a
device that processes electronic data from registers and/or memory to
transform that
electronic data into other electronic data that may be stored in registers
and/or memory. A
"computing platform" may comprise one or more processors. As used herein,
"software"
processes may include, for example, software and/or hardware entities that
perform work
over time, such as tasks, threads, and intelligent agents. Also, each process
may refer to
multiple processes, for carrying out instructions in sequence or in parallel,
continuously or
intermittently. The terms "system" and "method" are used herein
interchangeably insofar as
the system may embody one or more methods and the methods may be considered as
a
system.
[331] While one or more embodiments of the invention have been described,
various
alterations, additions, permutations and equivalents thereof are included
within the scope of
the invention.
[332] In the description of embodiments, reference is made to the accompanying
drawings
that form a part hereof, which show by way of illustration specific
embodiments of the
claimed subject matter. It is to be understood that other embodiments may be
used and that
changes or alterations, such as structural changes, may be made. Such
embodiments, changes
or alterations are not necessarily departures from the scope with respect to
the intended
claimed subject matter. While the steps herein may be presented in a certain
order, in some
cases the ordering may be changed so that certain inputs are provided at
different times or in
a different order without changing the function of the systems and methods
described. The
disclosed procedures could also be executed in different orders. Additionally,
various
computations that are herein need not be performed in the order disclosed, and
other
embodiments using alternative orderings of the computations could be readily
implemented.
In addition to being reordered, the computations could also be decomposed into
sub-
computations with the same results.
73
