A New Paradigm in Quantitative Investing

Nov 27 2007 | 10:51am ET

By Dimitri A. Sogoloff, President & CEO, Horton Point

Introduction: A few weeks after the unusually large drawdowns attracted everyone’s attention to the perils of “quantitative” investing, the popular opinion of what may have happened, has been formed. The answer, apparently, lies in the quantitative space becoming overcrowded, with most models generating similar forecasts and thus similar portfolios. Losses began with the rapid unwind of a large market-neutral equity portfolio and have generated ripples throughout the quant world.

Still un-answered is the more fundamental question: why have different quant groups, using supposedly different investment tools, ended up with similar models

The empirical evidence suggests that the majority of the equity market-neutral models are indeed similar in their approach to forecasting price behavior. In fact, they use one form of statistical analysis or the other (hence the commonly used term, “Statistical Arbitrage”). The first problem with statistical analysis of financial data is that it assumes stable relationships between market factors, which we know not to be the case. The other problem is that everyone is looking at the same data sets (after all, every security generates only one time series of historical returns).

This paper suggest an alternative approach to development of quantitative investment strategies; the one which eschews statistics in favor of more dynamic sciences (e.g. physics), and postulates that the future of quantitative investing lies in continuous scientific innovation and applications of modern scientific principles to capital markets.

A Bridge Between Science And Finance

A flirtatious relationship between the physical sciences and economics has existed for well over a century. In 1838, the French mathematician and economist Antoine Augustin Cournot proposed a bridge between the social sciences and mathematics in his publication, Recherchés. Later, in the 1870s, W. Stanley Jevons and Alfred Marshall, among others, borrowed ideas from mid-nineteenth century physics built on concepts from thermodynamics to develop fundamental economic ideas, such as Marginal Utility.

Paradoxically, economics and finance have not drawn extensively on the vast body of modern physical knowledge since that time.

For example, since the late 1980’s physicists have recognized that unpredictable time series are in fact not random and can be analyzed. Specifically, Chaos theory states that things which appear random to the naked eye may in fact arise from a complex but well-defined system.

These results, observed in electrical circuits, lasers, chemical reactions, biological systems and mechanical devices, triggered an interest in economic systems. Extensive theoretical and empirical studies have shown that the evolution of asset prices in financial markets might indeed be due to underlying nonlinear deterministic dynamics of several variables!

Many industries outside of the financial sector have subscribed to this paradigm and have extracted significant value from its practical applications. During the past few decades, physicists have achieved important results in the field of statistical mechanics, nonlinear dynamics and disordered systems. Today, these theoretical results are successfully implemented in fields like telephony, networking, bioengineering, pattern recognition, space technology, and weather forecasting.

Financial markets, similarly, can be viewed as open systems in which many small parts interact nonlinearly in the presence of feedback. However, no meaningful discoveries have been made in the world of finance in spite of the critical importance these theories hold in our understanding of the behavior of capital markets.

Only recently has scientific research’s influence on the field of finance become somewhat less episodic. Today terms like “fat tails”, “Garch”, “jump diffusion”, “clustering”, etc., are well accepted by finance professionals. Models with origins in physics, such as Monte Carlo simulations, stable Lévy processes, Markov chains, and Extreme Value Theory are successfully implemented and widely used in derivative modeling, risk management and event forecasting.

In fact, the links between finance and thermodynamics, molecular physics, artificial intelligence, mathematical linguistics, statistical mechanics and many other scientific disciplines are stronger than previously thought. Unfortunately, economists’ use of available scientific advances from other fields typically lags by decades (witness the widespread use of Statistical Arbitrage today).

By the time economics and finance embrace a scientific paradigm or model, scientists have already moved past them to different and more refined paradigms. This vast gap, between scientific idea and economic implementation, still exists and holds significant opportunities for those who are capable of systematically exploring the probabilities that discoveries in modern science may create in the financial markets.

Interdisciplinary Science

Historically, the most important technological creations emerged when two or more seemingly unrelated sciences opened dialogue with one another, sharing their histories, discoveries, and insights. These “bridges” were responsible for space travel, new drug discovery, earthquake forecasting, information technology, bioinformatics, the Internet, nanotechnology and wireless communication, to name a few.

Today, we reap the sophisticated scientific benefits of the brightest minds of the 1960s and 1970s. If electrical engineers, computer scientists, and electromagnetic physicists hadn’t united their efforts in cutting-edge research, we wouldn’t currently be using such innovative computer networks. More recently, if applied mathematicians and psychologists hadn’t envisioned a new frontier for cognitive science, we wouldn’t have made such important strides in language processing and brain imaging. If computer science, engineering, psychology, and neuroscience had never conversed, artificial intelligence wouldn’t have been one of the major scientific achievements of the late 20th century.

When an applied mathematician by the name of Fischer Black met an economist Myron Scholes, the option valuation formula was born and forever changed the way traders valued equity derivatives. This seminal work was recognized by the Nobel Committee in 1997. Similarly, the marriage of psychology and finance gave birth to a new science, behavioral finance. So important was this new hybrid discipline that the 2002 Nobel Prize in Economics went to its pioneer, Daniel Kahneman.

Despite these wonderful achievements, few practitioners in the field of finance have consistently used latest scientific advances to their advantage. There are a number of investment banks and multi-strategy hedge funds who actively use the work of the “quants” to optimize value within their traders’ portfolios. However, most traders do not have the time or desire to venture outside of their comfort zones, therefore relegating quants to supporting roles where they are not given the freedom to apply their ideas and theories directly to the investment process.

In investment management, the term “Quantitative Investment” often means using simple rule-based techniques which have shown limited success in the past.. Many systematic investment strategies rely on a limited set of historical statistics and simplistic rules rather than on a deeper understanding of the complex, dynamic, and non-linear relationships occurring below the surface.

The first practitioners in the algorithmic strategy space were CTAs (“commodity trading advisors”) who typically used statistical trend-following algorithms to make investment decisions.

Over the years, the number of these algorithms proliferated the industry and became easily accessible to the general investing public through numerous off-the-shelf software and “how-to” books, thereby rendering them obsolete.

At the same time, tremendously exciting horizons have opened to investors who take the science of finance seriously. Essentially, primary scientific research – the most expensive and time consuming work – is there for the taking. The diversity of scientific expertise today and the rate of its potential growth in the 21st century are extraordinary. It simply takes a good “bridge builder” to begin connecting the dots between the worlds of finance and science.

The Scientific Platform

Synthesizing these two worlds is the real challenge. It takes talented scientists to recognize similarities between their areas of expertise and the multifaceted world of capital markets. It also requires an ample amount of innovation and the desire and ability to think outside the box.

A pipeline of ideas generated by a “quantitative finance think tank” can be quite daunting considering the diversity of topics. Therefore, a sophisticated infrastructure is needed which encourages both intellectual freedom and the development of practical applications. This is a critical part of bridging the gap between theoretical research and practical benefits.

Large and successful interdisciplinary research organizations like Bell Labs, DARPA (Defense Advanced Research Projects Agency) and a few others have long recognized the value of a “unified scientific platform”. Such platforms included processes which took a scientist’s idea from theory to product. The time has come to begin thinking about such platforms in the world of capital markets.

The Science of Alternative Investments

The alternative investment industry is changing in front of our eyes. A large part of the industry is fast converging with more traditional asset classes. Many formerly “alternative” strategies are now subject to easy and accurate passive replication, and thus can no longer be considered alternative.

At the boundaries, the alternative investment industry continues to provide sufficiently uncorrelated returns to justify the high fees and lack of transparency. New “alternative alternatives” are developing in brand new markets – weather, power, emissions, insurance, etc.

“Alpha generation through innovation” should be the ongoing mantra for all alternative investment businesses. In our view, such innovation can be very rewarding when modern scientific knowledge is applied to solving complex non-linear problems that exist in the world of finance.

A new paradigm is waiting to be developed. Such a paradigm must address the need for optimal asset allocation, continuous creation, rotation and updating of diverse investment strategies. Only five years ago, this would not be possible. In recent years, many limitations of technology and computational capabilities have been lifted which opens up an entire new range of opportunities for generating the alternative, algorithmic, Alpha. In contrast with more popular rule-based market neutral or trend strategies, such Algorithmic Alpha would be a product of the truly innovative, repeatable, uncorrelated and dynamically adaptable investment process.

Dimitri A. Sogoloff is president and CEO of Horton Point. Horton Point conducts research, analysis, and management of next generation quantitative investment strategies combined with the thoughts, ideas and advancements from other areas of human progress. Through this forward-thinking and reflective model, the firm aims to capture a new generation of investment returns, which it calls Algorithmic Alpha. For more information, please contact: info@hortonpoint.com, (212) 939-7300.

If you would like to become a guest contributor, please CONTACT US

In Depth

AIMA: Smaller Firms Remain the Lifeblood of the Hedge Fund Industry

Jul 26 2017 | 5:55pm ET

It is a hedge fund industry truism that the largest managers receive the most attention...


CFA Institute To Add Computer Science To Exam Curriculum

May 24 2017 | 9:25pm ET

Starting in 2019, financial industry executives sitting for the coveted Chartered...

Guest Contributor

Rastegar: PE Real Estate Gains Momentum as Uncertainty Rises

Jul 21 2017 | 6:04pm ET

The steady march of equity markets and fundamental shift in the direction of Fed...


From the current issue of