ArticlePDF Available

A Dynamic Future for Active Quant Investing



Active quantitative portfolio management is on the verge of change, we believe towards a more flexible approach capable of capturing dynamics in risk and return expectations across an array of asset classes. The static quant-driven approach to active management in widespread use today is ill-equipped to deal with market environments that diverge substantially from typical conditions. We discuss what changes are needed at this important juncture for the active quant community to maintain relevance and improve the odds of long-term investment success. Among our recommendations, we suggest active quants broaden their focus by adopting a top-down (macro-driven) approach with the design flexibility to accommodate investment success in our complex dynamic capital markets.
Electronic copy available at: copy available at:
A Dynamic Future for Active Quant Investing
This is the submitted version of the following article:
A Dynamic Future for Active Quant Investing
By Xi Li and Rodney N. Sullivan
Journal of Portfolio Management, vol. 37, no. 3, Spring 2011 (p. 29-36)
Copyright © 2011, Institutional Investor, Inc.
Published in final form at:
Xi Li
XL Partners
Boston College
Rodney N. Sullivan
CFA Institute
We believe that active quantitative portfolio management is on the verge of
moving toward a more flexible approach capable of capturing dynamics in risk and
return expectations across an array of asset classes. In widespread use today, the
static quant-driven approach to active management is ill equipped to deal with market
environments that diverge substantially from typical conditions. We discuss what
changes are needed at this important juncture for the active quant community to
maintain relevance and improve the odds of long-term investment success. Active
quants must broaden their focus beyond typical systematic bottom-up quantitative
inputs to emphasize top-down qualitative evidence. Such a judgment-based, top-down
(macro-driven) approach offers the flexibility necessary to achieve investment success
in our complex, dynamic capital markets.
Acknowledgments: We are grateful for helpful comments from Paul Bukowski,
Yunfang Cai, Max Golts, Antti Ilmanen, James Picerno, Larry Siegel, Apurv Jain, Peter
Wallison, and James Xiong.
Electronic copy available at: copy available at:
A Dynamic Future for Active Quant Investing
Recent events have illuminated the structural problems of traditional active
quantitative management. During the recent global financial crisis (GFC), managers
struggled with the systematic quant approach, which resulted in portfolio
underperformance and unexpected losses to investorsrevealing major shortcomings
that have existed for years but now have become readily apparent. The standard
systematic models in today’s quant community are clearly ill equipped to handle
macroeconomic and investment environments that deviate substantially from typical,
long-run conditions. Even though the standard models are now being refurbished to
some degree, our experience suggests that a sea change, not marginal progress, is
needed to accommodate the reality of markets. Unfortunately, mainstream (not all)
active quant management remains focused on models that are inadequate for today’s
complex, adaptive markets.
Investors and financial economists have generally done a poor job of
understanding and incorporating the connection between capital markets and the
macroeconomy. Risk management and portfolio construction techniques must evolve
to accommodate a wider array of possible outcomes, to illuminate the changing nature
of our dynamic global capital markets, and to help guide and manage our decisions
meaningfully. Expert investment management requires a broad and informed view
across global markets. Although we are critical of certain ways that quant modeling is
often used today, our experience suggests that at this important juncture, the practice
of active quant portfolio management should not be abandoned altogether. Rather, we
propose an eclectic approach that combines quantitative modeling with informed
qualitative assessment and offers a robust technique for successful investing.
The inflection point at which we now find ourselves calls for a sharp refocusing
on effective approaches and tossing aside ineffective, “dilequant”1
1We borrowed this term from Mark Kritzman, who has applied it to those who merely dabble in
quant methods.
methods. A more
constructive way forward is within our grasp. Success requires a move toward a
dynamic, top-down (macro-driven) approach, one capable of capturing shifts in risk
and return expectations globally across an array of asset classes and market
Electronic copy available at: copy available at:
This article stems from our previous work at the intersection of investment
management and complex adaptive markets (X. Li 2007; Sullivan 2008, 2009a, 2009b,
2010; Sullivan, Peterson, and Waltenbaugh 2010). Expanding our focus beyond
quantitative inputs to emphasize empirically driven judgment, we examine portfolio
construction that optimizes the strength of signals by quantifying and qualifying
macro-oriented insights combined with solid, bottom-up security selection. Although
data are lacking, the active quant community does not appear to be pursuing our
suggested, macro-driven approach.
Active Alpha
Especially in recent years, the development of modern portfolio theory has
provided the backdrop for the widespread acceptance of the techniques of active
quants and the rise in their popularity. Like all successful investors, successful active
quants make decisions on the basis of informed theory and keen historical insight,
thus fostering market efficiency and sustainability. Although the active quant
approach is difficult to ascertain, its primary component is a systematic, structured
investment process that involves the large-scale collection of relevant current and
historical information summarized in a meaningful, nonarbitrary way for use in
making decisions about the future.2
In essence, active quants seek superior performance through the gathering and
distilling of complex historical information in a systematic fashion.
Active quant management is typified by the
structured, system-driven approaches of Grinold and Kahn (2000).
In their purist form, active quant factor models attempt to systematically
capture recurring deviations from the general equilibrium CAPM market model. Active
Quants are so
called because of their intense focus on “quantifying” unique alpha insightsthose not
already priced into the market. This quantifying is accomplished via formulaic
empirical factor models applied across a wide array of assets in a hunt for statistically
significant deviations from historical trends. The basic idea is successful forecasting
driven by identifying signals (skill) and applying that skill as often as possible per year
2We use the terms active quant, active alpha, and active portfolio management synonymously.
By quant, we mean traditional active portfolio management strategies, not other quant forms,
such as statistical arbitrage or high-frequency trading.
3Of course, this definition is not limited to quant investors. For a brief overview of both
fundamental and quant investing, including pitfalls, see Sullivan (2010).
quants view the traditional CAPM, as driven by the single market factor, an
oversimplification of the realities of markets. Active quants thus seek to generalize the
traditional CAPM with skillful exposures to additional tradable factors. Though there
exists no predetermined set of factors, many strategies and factors do exist. Examples
of such insights include the various documented behavioral biases exhibited by
investors. Collectively, this set of factors forms the driving force behind abnormal
returns and can span a wide range of likely alpha sources. We can easily understand
how investors view the culture of active portfolio management as a valuable and
disciplined process for extracting alpha. By our very nature, we humans love
structure. It is elegant in its intuitive appeal and implementation.
A distinct advantage touted by active quant practitioners is that the process
leads to a highly disciplined investment approach: A rigorous, variable-centric
approach that follows a logical, hypothesis-oriented decision process is largely
uninhibited by the various cognitive lapses in judgment exemplified by the average
investor. Furthermore, computer models have the superior ability to process vast
amounts of information rapidly. Given the dynamic complexity of markets and the
many assets at play, active quant modeling is undoubtedly a useful aid in the ongoing
search for alpha. These various quant attributes can be turned into a competitive
advantage when applied in the right circumstances. Before discussing this way
forward, let us turn our attention briefly to the ways that quant models frequently trip
us up in our search for performance.
Active Alpha Shortcomings
Any discussion of the challenges of successful active portfolio management
must begin with the fact that alpha sources are very rare, even rarer than commonly
supposed. To see why, let us consider persistent alpha capture, which, as noted by
Jarrow (2010), requires two conditions: (1) The activities of arbitrageurs do not
eliminate market mispricing and (2) the alpha source is continually funded in a
structural way, either knowingly or unknowingly, thereby allowing arbitrageurs to
reap profits. Of course, such market conditions occasionally exist, but given these two
rather stringent conditions, alpha is certainly not present in abundance.4
Now that we know our biggest challenge is finding active alpha, let us briefly
consider the various ways in which the process-driven active quant approach, as it is
sometimes applied today, can hinder success. Key obstacles can be found in the model
development process: the systematic sourcing and implementing of the active signals.
These issues combine to give active quant investors a false sense of skill.
For this and
other reasons discussed later, active alpha is more elusive than many believe.
The most pernicious issue concerns the inability of static active quant models
to capture change or the unexpected effectively. The future direction of marketslike
the path of a bird just released in flightis highly uncertain. Although computer
speed and mathematical models are a powerful combination in ferreting out vast
amounts of information and recurring patterns, automating human nature effectively
remains elusive. As recent events have demonstrated, markets seem to mock the
structured, disciplined approach of models. They abruptly jump from one equilibrium
point to another, giving yield to fat tails and rendering the commonly used static
structural models impotent. Consider how, during the GFC, governments around the
world implemented a flood of unconventional policies at lightning speed. These shocks
quickly introduced a variety of new elements that influenced financial markets in
unprecedented ways. No single model can contain all the information necessary to
capture the uncertain path of outcomes and their attendant consequences. A more
dynamic approach, accompanied by empirical judgment, is needed.
Unfortunately, many active approaches have been oversimplified. For example,
they are based on the assumptions of normal distribution and stationarity. That is,
market prices will behave as they generally have in the recent past with the hope that
the strategy can be adjusted just before markets start to change direction. Over a
series of market cycles, little or no value is added by such approachesindeed,
negative value add is most often the result after fees.5
4One example of persistent alpha is the value of the call option implied by a residential
mortgage-backed security (RMBS) mortgage, which has historically tended to be overpriced by
homebuyers, thus affording RMBS buyers an arbitrage opportunity. Will this source remain
viable in the future?
5 For example, Agarwal and Naik (2004) and Mitchell and Pulvino (2001) find that many hedge fund strategies
possess characteristics similar to a put option.
Of course, history serves as an important guide to understanding the
interconnectedness of markets and the range of possible outcomes. In reality, market
linkages, even those established over long horizons, are dynamic, subject to the ebbs
and flows of markets. Models that are “overfittedon the basis of historical dynamics
can be severely hampered by the structural changes and sudden stops that financial
markets frequently experience.6
Active quant managers attempt to deal with overfitting in a variety of ways,
some of which may be useful and effective tools in the search for alpha (e.g., regime
switching, extreme value theory), but other methods are downright problematic (e.g.,
winsorization, which essentially ignores extreme events and volatile markets
altogether). Perhaps well-established relationships will revert to their old ways
following a particular event, or maybe the newly established structural relationships
will become dominant. In any case, uncertainty cannot be removed from the equation.
Factors are stochastic and time-varying processes; they are inadequate
representations of the reality of complex, adaptive, and unpredictable markets.
These changes can significantly alter the relationships
established by a model temporarily or even permanently.
Put another way, static active alpha factors are built to perform well over time,
on average. Success thus depends on the future’s mimicking history in a structured,
persistent way. But history repeats itself only in broad tendencies. Perhaps driven, in
part, by physics envy,many active quants treat factors as factual and permanent,
when they should be viewed as merely temporary, as suggested by the social nature of
markets. Guided by the free will of agents, markets and the economy adapt
responsively over time. Using the same overfitted model parameter weightings through
all macroeconomic cycles will likely not work well. In short, static alpha factors fitted
with historical data may prove to be false factors when applied to future horizons.
Another not-so-subtle point is that we cannot know the true general
equilibrium model that describes market returns. Models miss the richness of
complex, adaptive markets. As stated earlier, although CAPM is a solid theory, it is
viewed as naive by modern investment managers and thus possesses limited
applicability. The well-known FamaFrench size and style factors represent a
6To varying degrees, all investors rely on historical information to guide their decisions.
Therefore, rigorous historical analysis is not necessarily problematic unless the user expects
the future to mirror the past. The backward-looking nature of quant investing, however, should
serve as a reminder of the limits of quant models.
constructive step forward in generalizing the CAPM. But variables are always missing
from any model that attempts to capture the interactions in capital markets. Even the
best statistical models make simplifying (or hidden) assumptions and can incorporate
only a small number of variables, thus obscuring the variables’ implications.
Even if we could accurately define the true market model, model builders can
describe only those aspects of financial markets that are measurable or can be
reasonably proxied (e.g., the value metric can be measured in many reasonable ways).
Some desirable inputs, however, cannot be reasonably measured or proxied. Consider
the modeling of the accrual anomaly associated with earnings gimmickry. Effective
arbitrage of the impact of any accrual-related earnings management requires an
estimate of those accruals for which management exercises discretion (normal
accruals are difficult to manipulate). Only imperfect proxies exist for discretionary
accruals, however, and any inputs are measured with error. X. Li (2010), for instance,
shows that discretionary accruals are just one aspect of earnings gimmickry. He goes
on to demonstrate how additional earnings management activities regularly used by
managers can also be used to predict performance. In sum, forecast error related to
the modeling process emanates from many sources: omitted or mismeasured
variables, misspecification, overfitting, and obscured or wrong proxies or factors. In
our experience, these errors have led to costly, unintentional bets for some investors.
The empirical weakness of common quant models is evidenced by weak
explanatory power (low R2s), accompanied by significant idiosyncratic volatility.
Residual risk imposes real uncertainty, and it is a primary factor in limiting both
effective arbitrage and the extraction of alpha. In turn, investors seeking to profit from
various anomalies must bear greater uncertainty in outcomes than was previously
understood (see Pontiff 1996, 2006; X. Li and Sullivan, 2011; X. Li and Sullivan, 2010;
and X. Li and Sullivan, 2011a). This uncertainty comes in the form of high
idiosyncratic risk, and transactions costs, which raises costs and hinders the
profitable arbitrage of seemingly anomalous effects. The foregoing leads us to conclude
that active quant models frequently possess biases in what modelers believe to be true
about the relationships established via the modeling process.
Moreover, as alpha opportunities become known, overcrowding occurs.7
The aforementioned issues also present major challenges to the business of
implementing the often dense and opaque quant models. Even though quants may
embrace Einstein’s advice to make everything as simple as possible but no simpler,
highly structured models geared to complex, dynamic markets are necessarily complex
themselves. Although investment personnel can grasp a model’s key insights, the
difficulty in gaining a full, working knowledge seems to increase with the square of the
number of inputs. Operational managers are thus challenged to ensure that all
segments of the firm fully understand the implications of any model put into practice.
For success, one must know what risks both the firm and its clients face in the
investment processand how to manage those risks effectivelyas well as what
changes are needed to improve (or worse, fix) a model that is not producing.
noted by Lo (2004), markets are not only complex but also adaptive. Such crowding
into the trade has decreased (or eliminated) the persistence of the well-known alpha
opportunities and may even result in a sudden reversal of well-established
relationships (e.g., large caps outperforming small caps for an extended period). For
instance, note that the equity market flare-up in early August 2007 was deemed to be
driven, in large part, by a sudden downdraft of concentrated equity holdings common
to many quant equity funds (Kandhani and Lo 2007). Increased trading commonality
also suggests a rise in market vulnerability as evidenced by increased systematic risk
in recent years (Xiong and Sullivan 2011).
For all these reasons, one may fairly say that these challenges create a high
degree of uncertainty for typical quant performance. Static models make little sense in
a world where risk premiums, correlations, and volatility change dramatically from
year to year.
As mentioned earlier, the rigorous decision-making process of active quants has
value under certain circumstances; but when misused or overused, it can be
detrimental. Models are merely a guide for quality decision making in the face of
uncertainty. They interpolate and extrapolate the world, but by no means offer the
correct interpretation of reality. Therefore, those depending on quant models have a
7Although nuances certainly exist, quant alpha frameworks bear a striking similarity. As
evidence, also consider how most quants rely on the same risk and optimization models
provided by a limited number of vendors, many of whom deliver data updates on the same
schedule. See Li (2011) for a discussion of the optimal global equity investment framework.
responsibility to be keenly aware of the limits, boundaries, and risks of quant
techniques and to know when to turn away and seek input elsewhere. This task
becomes more challenging as finance grows increasingly complex.
The shortcomings of active alpha offer a compelling reason to dispense with
ineffective approaches to active quantitative management and refocus attention on the
ways that modeling can be effective. In this way, quants can gain a competitive edge
while supporting client goals.
The Future of Active Quant Investing
Active portfolio management is at an inflection point. The recent market
turbulence has reinforced the notion that active managers are not worth their high
fees. The active quant community has seemingly suffered more than most. Adding to
these troubles, investors are increasingly turning to passive and semi-passive, low-
cost index funds. We predict that such index strategies (in particular, in the form of
exchange-traded funds) will continue to absorb market share in the coming years,
especially from the active quant side, because such strategies in practice today are
easily replicated in a growing roster of fund offerings.
These issues have certainly not gone unnoticed by the active quant community,
which continues to try to improve the effectiveness and efficiency of traditional quant
models, mostly through active signal innovation. Although likely to meet with some
success, this additive approach, in our view, does not go far enough. As discussed
earlier, adhering to a set of fixed factor weights (established as the average of various
market cycles) will lead to unacceptable performance periods. Designing flexible
models with an eye toward quantifying big-picture issues warrants considerably more
attention (see Exhibit 1). Model construction should thus emphasize empirically
based evidence that includes specifications to appropriate risks and the nimbleness to
cope with market turbulence. Models should work well in circumstances that matter.
Exhibit 1
Macro–Risk Factors
Size/Style Rotation
Economic Growth
The myriad of challenges facing the active quant community notwithstanding,
we foresee a dynamic future for active quant investing. This (not altogether) novel
breed of active quant asset management will exhibit an interdisciplinary culture
focused on extending and updating (not abandoning) the existing core competencies
held by the typical active quant firm. In light of the challenges to achieving investment
success, the need for an agile, top-down, active asset management approach that
integrates both quantitative and fundamental insights is clear. Managers should take
risks if the returns appear to represent fair compensation. At any given time, some
asset classes may offer an acceptable or even generous compensation while others
may offer an unacceptable trade-off. The idea is to take full advantage of time-varying
risk premiums, driven, in large part, by investors’ cycling between risk aversion and
risk adoration.
On this last point, capturing risk premiums cannot be accomplished with
overfitted static models developed via back testing. A forward-looking orientation will
emphasize an eclectic framework with the flexibility to capture the top-down risk
dynamics of markets. In practice, this approach means broadening the inputs to
include global macroeconomic views, which will play a more prominent role in asset
management. The prevailing framework thus switches from static to dynamic asset
allocation and takes into account time-varying macroeconomic forecasts that are
overridden by qualitative judgment, as necessary. With such skill comes the ability to
recognize when a new structural cycle has begun.
In other words, successful macrofactor timing requires the transmission of
informed, empirically based judgment into the portfolio. Moving away from a narrow
focus (e.g., on the convergence to the norm of company-specific factors) and toward
broader macrofactors can be achieved only through a more qualitative approach. The
weak statistical significance of many macrovariables may give pause to some in
adopting macrovariables into the formal model framework. But given the frequent,
systemic, turbulent events affecting markets, such insights may indeed be highly
economically significant. They should certainly not be ignored. A judgment-oriented
approach based on empirical evidence will enable investors to better guide risk taking
and thus help clients succeed over the long haul. Obtaining a statistically significant,
but economically insignificant, alpha when the total portfolio is substantially down
does little to help investors achieve their goals.8
Dynamic global asset allocation will necessarily be complemented with a
rigorous, bottom-up assessment of assets, as always. In this connection, an innovative
strategy may well set aside parsimony in favor of a looser, more qualitative approach
for example, one that mixes fundamental with active quant investing and that
integrates bottom-up stock selection with top-down global macroinvesting. Such a
rich, integrated process that explores the full range of possibilities could offer the best
chance of performance successsay, in recognizing when a new world order is upon
us. Active investment managers could turn macro-driven market cycles into a
competitive advantageadmittedly a delicate task, calling for a competency in
dynamic global asset allocation.
Our recommended framework is consistent with Xiong, Ibbotson, Idzorek, and
Chen (2010), who found that about 80 percent of total return variation is dominated
by changes in the general market, with the remainder evenly split between specific
asset allocation and security selection. In other words, passive asset allocation and
active asset allocation together account for the bulk of return variability. Our
suggested approach is also consistent with Markowitz (2005), who found that the
market portfolio is not necessarily an efficient portfolio.9
Our top-down approach also places dynamic risk management at the fore.
Attention will be paid to much more than market risk. Investors must also give ample
attention to liquidity risk, counterparty risk, systemic risk, and the effects of leverage
8 Several statistical issues also inhibit the effectiveness of pure top-down quant approaches. First, modeling extreme
events yields very small sample sizes and thus little chance at statistical significance even when estimated over a
long historical sample. Second, return modeling generally requires fixed intervals such as monthly, quarterly, or
annual intervals. Such fixed modeling intervals cannot successfully capture the turning points of market cycles.
Maybe for this reason, NBER can only mark the end of a recession long after it has ended.
9Specifically, Markowitz (2005) suggested that the inefficiency of the market portfolio could be
so substantial that it would not be arbitraged away even if some investors could borrow
without limit.
This idea appears to be gaining traction. Witness the recent surge of interest in
gaining a better understanding of fat tails (which imply a more frequent occurrence of
extreme events), dynamic correlations, and systemic risk (see, e.g., Kritzman and Y. Li
2010; Sullivan et al. 2010; Xiong and Idzorek 2011). These models provide a
framework that illuminates the changing nature of risk and helps guide and manage
risk decisions.
In recent years, investors have turned their attention to greater diversification
of asset classes in order to generate higher returns and protect assets from market
volatility. We certainly agree that a broad array of asset classes can improve portfolio
efficiency and should thus be included in portfolios. This notion is consistent with
Tobin (1958) and Sharpe (1964), who suggested that portfolios be formed with a
combination of the risk-free asset and the market portfolio of all risky assets.
Nevertheless, many investors undertook asset allocation under the false premise that
certain asset classes would yield equity-like premiums while being largely
uncorrelated with the overall market. But, when evaluated carefully in realityand as
many learned painfully during the GFCthe uncorrelated asset class is a myth (Ennis
2009). The lesson here is simple: Collecting a risk premium means that you must bear
the riskthere is no free lunch in asset allocation.
Altogether, this suggests that the primary focus should be on that which plays
the biggest role: the asset allocation decisionthat is, the process of deciding how best
to adjust asset allocation in accordance with a sensible assessment of the riskreturn
trade-off that the various markets are offering. Again, the decision to investwhether
to take riskis the most important investment decision. How much market risk to
take entails the two aspects of asset allocation (policy allocation to the general market
and specific asset allocation). The powerful intuition behind our approach is that
proper portfolio construction is an ongoing, dynamic process, one of calibrating from
the top down the set of risk and return expectations for each asset (asset class,
country/region, industry, and security) against current and expected macroeconomic
and investment conditions. Although challenges remain, the upside is a compelling,
flexible framework, one that integrates the time-varying macropicture with
microspecific analysis. Such an approach is within our grasp.
Although obstacles to implementing our suggested approach to portfolio choice
undoubtedly exist, the evidence makes clear that successful investing is dominated by
value-driven active asset allocation. As Peter Bernstein (2003) deftly observed, the
traditional strategic approach of fixed-asset allocation is outmoded. The challenge of
portfolio choice is much more than merely selecting for inclusion uncorrelated asset
classes that constitute significant economic exposure and then specifying a fixed
proportion of each.10
As noted by Sharpe (2010), the only portfolio that all investors can hold
simultaneously is the market-weighted portfolio of all assets. So, holding the market
portfolio is best for those without superior knowledge about the relative attractiveness
of asset classes. Anything else is an active strategy. This observation plainly suggests
that the traditional approach of rebalancing to prespecified weights represents an
active contrarian strategy. But, as we have shown, no single fixed-weight strategic
asset allocation is best for all environments.
For all these reasons, active quantitative investment management appears to be
on the verge of important change, which is not necessarily a bad thing. At the very
least, it suggests an opportunity. An approach that incorporates a nimble and
dynamic global asset allocation is more likely to yield positive results for all interests
(both principals and agents) over the long haul. Although ours is perhaps not an
entirely novel approach, it is certainly not the current dominant one.
This discussion leads us back to our
main thesis: An industry shift is needed, with a focus on the capacity to understand
the trade-off between asset classes from a macropoint of view—an eclectic approach
that adapts to the market situation. With skillful insight and wide breadth, algorithms
can go beyond a systematic narrative of what has merely happened in the past. This
approach means implementing all relevant quantifiable and qualitative factors,
whether or not those factors can be systematically incorporated.
Active equity management has become increasingly narrow and complex, but
not necessarily wiser. As a result, investors will likely continue to embrace passive and
semi-passive low-cost index strategies and shun high-fee, hidden-beta strategies. For
these reasons, we are at an important juncture in active quantitative investment
10For further discussion on this topic, see Sullivan (2008).
11Of course, managers should regularly compare their asset allocations with current market
management. To navigate this juncture successfully, the skills of quants can be
tapped for a competitive advantage, but only in the right circumstances.
We envision a flexible and nimble investment approach, one that will more
likely deliver performance success over both the intermediate term and the long haul.
We see quantitative asset management turning to an eclectic framework that
accommodates a wider array of possible outcomes and that copes with the frequent
occurrence of extreme events (fat tails). The distinguishing feature of our technique is
the recognition that investors must negotiate turbulent periods. Todays standard
active quant models that use static linear methods will, by their very nature, prove
inadequate in such markets. Therefore, expert models must dynamically reflect all
portfolio risk exposures, not merely those represented by typical conditions and
captured in static models. Qualitative judgment based on empirical evidence must
meaningfully accompany any quant-driven, decision-making process.
In sum, we call on the active quant investment community to broaden its focus
beyond the standard bottom-up, systematic model by incorporating a dynamic, top-
down (macro-driven) approach to investment management, one with the flexibility to
capture shifts in risk and return expectations across an array of asset classes and
market environments. Quant methods can be highly useful when accompanied by
qualitative reasoning. Given the extreme events that markets frequently experience,
modelers must take seriously their responsibility to engage deeply.
Endnote: This article represents the views of the authors and does not represent the
official views of the authors’ employers.
Bernstein, Peter L. 2003. “Are Policy Portfolios Obsolete?” Economics and Portfolio
Strategy (March 1).
Ennis, Richard M. 2009. “The Uncorrelated Return Myth.” Financial Analysts Journal,
vol. 65, no. 6 (November/December):6–7.
Grinold, Richard C., and Ronald N. Kahn. 2000. Active Portfolio Management: A
Quantitative Approach for Producing Superior Returns and Controlling Risk. 2nd ed.
New York: McGraw-Hill.
Jarrow, Robert A. 2010. “Active Portfolio Management and Positive Alphas: Fact or
Fantasy?” Journal of Portfolio Management, vol. 36, no. 4 (Summer):1722.
Kandhani, Amir E., and Andrew W. Lo. 2007. “What Happened to the Quants in
August 2007?” Journal of Investment Management, vol. 5, no. 4:5–54.
Kritzman, Mark, and Yuanzhen Li. 2010. “Skulls, Financial Turbulence, and Risk
Management.” Financial Analysts Journal, vol. 66, no. 5 (September/October):30
Li, Xi. 2007. “Unintentional Bets of Pure Quant Investing.” Working paper.
Li, Xi. 2010. “Real Earnings Management and Subsequent Stock Returns.” Working
———. 2011. “Assessing Alternative Global Equity Investment Frameworks.” Working
Li, Xi, and Rodney N. Sullivan. 2010. “The Limits to Arbitrage: Why Low-Risk Stocks
Outperform.” Working Paper.
Li, Xi, and Rodney N. Sullivan. Forthcoming 2011.The Limits to Arbitrage Revisited:
The Accrual and Asset Growth Anomalies.” Financial Analysts Journal, vol. 67, no.
4 (July/August).
Li, Xi, and Rodney N. Sullivan. 2011a.The Limits to Arbitrage Revisited: International
Evidence for the Asset Growth Anomaly.” Working Paper.
Lo, Andrew W. 2004. “The Adaptive Markets Hypothesis.” Journal of Portfolio
Management, 30th Anniversary Issue (Autumn):1529.
Markowitz, Harry M. 2005. “Market Efficiency: A Theoretical Distinction and So
What?” Financial Analysts Journal, vol. 61, no. 5 (September/October):1730.
Pontiff, Jeffrey. 1996. “Costly Arbitrage: Evidence from Closed-End Funds.” Quarterly
Journal of Economics, vol. 111 (November):11351151.
———. 2006. “Costly Arbitrage and the Myth of Idiosyncratic Risk.” Journal of
Accounting and Economics, vol. 42:3552.
Sharpe, William F. 1964. “Capital Asset Prices: A Theory of Market Equilibrium under
Conditions of Risk.” Journal of Finance, vol. 19, no. 3 (September):425442.
———. 2010. “Adaptive Asset Allocation Policies.Financial Analysts Journal, vol. 66,
no. 3 (May/June):4559.
Sullivan, Rodney N. 2008. “Taming Global Village Risk.” Journal of Portfolio
Management, vol. 34, no. 4 (Summer):5867.
———. 2009a. “Governance: Travel and Destinations.” Financial Analysts Journal, vol.
65, no. 4 (July/August):6–10.
———. 2009b. “Taming Global Village Risk II: Understanding and Mitigating Bubbles.”
Journal of Portfolio Management, vol. 35, no. 4 (Summer):131141.
———. 2010. “Competence Trumps Style.” Financial Analysts Journal, vol. 66, no. 1
Sullivan, Rodney N., Steven P. Peterson, and David T. Waltenbaugh. 2010. “Measuring
Global Systemic Risk: What Are Markets Saying about Risk?” Journal of Portfolio
Management, vol. 37, no. 1 (Fall):6777.
Tobin, James. 1958. “Toward a Theory of Market Value of Risky Assets.” Review of
Economic Studies, vol. 25, no. 2:6586.
Xiong, James X., and Thomas M. Idzorek. 2011. The Impact of Skewness and Fat
Tails on the Asset Allocation Decision.” Financial Analysts Journal, vol. 67, no. 2
Xiong, James X., Roger G. Ibbotson, Thomas M. Idzorek, and Peng Chen. 2010. “The
Equal Importance of Asset Allocation and Active Management.” Financial Analysts
Journal, vol. 66, no. 2 (March/April):2230.
Xiong, James X. and Rodney N. Sullivan. How the Rise in Passive Investing
Contributes to Market Vulnerability.” Working paper. 2011.
... The shortcomings of MPT highlighted above should be seen in light of the fact that investors are now shifting to a more flexible approach to portfolio optimisation, one that can capture the dynamics in risk and return expectations across an array of asset classes. Li and Sullivan (2011) suggest that the traditional strategic approach of fixed asset allocation is outdated. ...
Full-text available
This paper highlights the shortfalls of Modern Portfolio Theory (MPT). Amongst other flaws, MPT assumes that returns are normally distributed, that correlations are linear and risks are symmetrical. We propose a dynamic and flexible scenario-based approach to portfolio selection that incorporates an investor's economic forecast. Extreme Value Theory (EVT) is used to capture the skewness and kurtosis inherent in asset class returns and account for the volatility clustering and extreme co-movements across asset classes. The estimation consists of using an asymmetric GJR-GARCH model to extract filtered residuals for each asset class return. Subsequently, a marginal cumulative distribution function (CDF) of each asset class is constructed by using a Gaussian-kernel estimation for the interior, together with a generalised Pareto distribution (GPD) for the upper and lower tails. The distribution of exceedance method is applied to find residuals in the tails. A Student's t copula is then fitted to the data to induce correlation between the simulated residuals of each asset class. A Monte Carlo technique is applied to simulate standardised residuals, which represent a univariate stochastic process when viewed in isolation but maintain the correlation induced by the copula. The results are mean-CVaR optimised portfolios, which are derived based on an investor's forward-looking expectation.
Alpha factors are built to perform well over time, on average. There are instances when they do not, and knowing these instances ex ante can be a significant source of added value for investors. The authors argue that factor failure is a function of its broad risk, and propose appropriate variables to measure it. They adopt a nonparametric model that predicts instances of likely factor failure, based on these variables, demonstrating that an implementable dynamic strategy based on our analysis generates a rewardtorisk ratio approximately four times that of a static approach, and about one and a half times that of an alternative dynamic approach based on momentum.
To exploit return predictability via dynamic asset allocation, investors face the important practical issue of how often to rebalance their portfolios. More frequent rebalancing uses statistically and economically significant short-horizon return predictability to aggressively pursue the dynamic investment opportunities afforded by changes in expected returns. However, the degree of return predictability typically appears stronger at longer horizons, which, along with lower transaction costs, favors less frequent rebalancing. The authors analyze the performance effects of rebalancing frequency in the context of dynamic portfolios constructed from monthly, quarterly, semi-annual, and annual return forecasts for U.S. stocks, bonds, and bills, where the dynamic portfolios rebalance at the same frequency as the forecast horizon. Along the transaction-cost/rebalancing frontier, monthly (annual) rebalancing provides the greatest outperformance when unit transaction costs are below (above) approximately 50 basis points, and dynamic portfolios based on annual rebalancing typically outperform the benchmarks for unit transaction costs well in excess of 400 basis points.
We propose a model of portfolio selection that adjusts an investors’ portfolio allocation in accordance with changing market liquidity environments and market conditions. We found that market liquidity provides a useful “leading indicator” in dynamic asset allocation. Specifically, market liquidity risk premium cycles anticipate economic and market cycles. Investors can therefore act to avoid markets with low liquidity premiums, waiting to extract liquidity risk premiums when the likelihood of extracting a liquidity premium improves. The result, meaningfully enhanced portfolio performance through economic and market cycles, and is robust to transactions costs and alternate specifications.
We present several active strategies for combining value and momentum strategies in a tactical asset allocation (TAA) framework. We refine the basic yield approach to valuation by standardizing the value signal using the Z-score. Such standardization not only enables us to directly compare valuation measures across asset classes, but also offers insight about each asset class’s absolute valuation by its own standard. Under the nonlinear approach, it helps to identify market peaks and bottoms. We improve the momentum strategy by considering both relative and absolute performances. In the combined tactical asset allocation model, this modification to momentum acts as a simple mechanism to adjust the importance of value and momentum strategies under different market conditions. Our combined model takes advantage of both short-term momentum effects and long-term mean-reversion in valuation to achieve superior overall portfolio performance. Finally, we also provide alternative models for smaller tracking errors.
Using idiosyncratic volatility as a proxy for arbitrage costs, the authors found that the highly publicized accrual and asset growth anomalies exist because of high barriers to arbitrage, occurring predominantly in the universe of stocks with higher arbitrage risks. Therefore, investors who seek to profit from the accrual and asset growth anomalies must bear greater uncertainty in outcomes than was previously understood.
We propose a unique dynamic portfolio construction framework that improves portfolio performance by adjusting asset allocation in accordance with a forecast of market risk. We find that modifying asset allocation according to our market risk barometer offers investors the promising opportunity to meaningfully enhance portfolio performance across market environments.
Full-text available
What is the relative importance of asset allocation policy versus active portfolio management in explaining variability in performance? Considerable confusion surrounds both time-series and cross-sectional regressions and the importance of asset allocation. Cross-sectional regressions naturally remove market movements; therefore, the cross-sectional results in the literature are equivalent to analyses of excess market returns even though the regressions were performed on total returns. In contrast, time-series analyses of total returns do not naturally remove market movements. Time-series analyses of excess market returns and cross-sectional analyses of either total or excess market returns, however, are consistent with each other. With market movements removed, asset allocation and active management are equally important in determining portfolio return differences within a peer group. Finally, an examination of period-by-period cross-sectional results reveals why researchers using the same regression technique can get widely different results.
Full-text available
The authors modeled the non-normal returns of multiple asset classes by using a multivariate truncated Lévy flight distribution and incorporating non-normal returns into the mean-conditional value at risk (M-CVaR) optimization framework. In a series of controlled optimizations, they found that both skewness and kurtosis affect the M-CVaR optimization and lead to substantially different allocations than do the traditional mean–variance optimizations. They also found that the M-CVaR optimization would have been beneficial during the 2008 financial crisis.
One of the most influential ideas in the past 30 years is the efficient markets hypothesis, the idea that market prices incorporate all information rationally and instantaneously. The emerging discipline of behavioral economics and finance has challenged the EMH, arguing that markets are not rational, but rather driven by fear and greed. Research in the cognitive neurosciences suggests these two perspectives are opposite sides of the same coin. An adaptive markets hypothesis that reconciles market efficiency with behavioral alternatives applies the principles of evolution-competition, adaptation, and natural selection-to financial interactions. Extending Simon's notion of "satisficing" with evolutionary dynamics, the author argues that much of what behaviorists cite as counter-examples to economic rationality-loss aversion, overconfidence, overreaction, mental accounting, and other behavioral biases-is in fact consistent with an evolutionary model of individual adaptation to a changing environment via simple heuristics. The adaptive markets hypothesis offers a number of surprisingly concrete implications for portfolio management.
With the aid of some simplifying assumptions, the capital asset pricing model comes to dramatic conclusions about practical matters, such as how to choose an investment portfolio and how to value financial assets. As illustrated in this article, when one particular, clearly unrealistic CAPM assumption is replaced by a more real-world version, some of the dramatic, practical conclusions of CAPM no longer follow. This result has implications for financial practice, research, and pedagogy.
It is commonly believed that active portfolio management can generate positive alphas.This is partly based on the belief that positive alphas represent disequilibrium returns, which can exist in complex financial markets.In contradiction, this article shows that positive alphas represent arbitrage opportunities, not just disequilibrium returns.As persistent and frequent arbitrage opportunities are much rarer, even in complex markets, Jarrow argues that positive alphas are more fantasy than fact. He introduces the notion of an unobservable factor that can generate false positive alphas, and which resolves the inconsistency between common belief and the sparsity of positive alphas.
A great many people provided comments on early versions of this paper which led to major improvements in the exposition. In addition to the referees, who were most helpful, the author wishes to express his appreciation to Dr. Harry Markowitz of the RAND Corporation, Professor Jack Hirshleifer of the University of California at Los Angeles, and to Professors Yoram Barzel, George Brabb, Bruce Johnson, Walter Oi and R. Haney Scott of the University of Washington.
This article proposes an asset allocation policy that adapts to market movements by taking into account changes in the outstanding market values of major asset classes. Such a policy considers important information, reduces or avoids contrarian behavior, and can be followed by a majority of investors.