1 / 30

What Can Quants Do Now?

What Can Quants Do Now?. The Intertek Group Tel: +33 1/45 75 51 74 Email: sfocardi@theintertekgroup.com www.theintertekgroup.com. Difficult environment…. A period of reduced profits and in some cases large losses Created by an unsustainable disequilibrium

vaughan
Download Presentation

What Can Quants Do Now?

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. What Can Quants Do Now? The Intertek Group Tel: +33 1/45 75 51 74 Email: sfocardi@theintertekgroup.com www.theintertekgroup.com The Intertek Group

  2. Difficult environment… • A period of reduced profits and in some cases large losses • Created by an unsustainable disequilibrium • Between economic fundamentals and financial flows (too much money chasing the same assets) • Plus herding phenomena of various sorts (same data, same models, same investment ideas) The Intertek Group

  3. Creates challenges for quants… • Traditional sources of alpha have disappeared: no stable trends, no stable spreads • New sources of profit change continuously: predictors change, momentum is intermittent, reversion to the mean unstable • Quantitative fundamentals are uncertain and not strongly correlated to returns The Intertek Group

  4. Key challenges to quants today • Identify new sources of profit: factors and predictors (fundamental characteristics of firms, patterns of prices and returns, …) • Understand how predictors change in time (changing correlations, changing feedbacks) • Understand the sustainability of the sources of profit • Understand systemic risk • Integrate new sources of information such as outcomes of quantitative text analysis The Intertek Group

  5. New tools for identifying proprietary factors with predictive power • New research shows: the signal-to-noise ratio determines factor uniqueness • If the signal-to-noise ratio were high, factors would be unique • But: in most large return universes the signal-to-noise ratio is low • Therefore: many different factor models can be defined (only a local approximation) The Intertek Group

  6. Non linearities • As linear factor models are only a local approximation, • Need to evaluate the goodness of each model in terms of forecasting or risk exposures • And to capture the non linearities • Without incurring the “curse of dimensionality ” problem • Due to the many parameters to estimate The Intertek Group

  7. Data/model compromise • Every factor model is a compromise between data-constrained estimation and model accuracy, • And captures only one specific aspect of returns • Hence: the need to evaluate what factors are key to providing good forecasts • And: anticipate when markets change The Intertek Group

  8. Borrowing from financial econometrics : the SNR The signal-to-noise ratio (SNR), a fundamental measure in many engineering disciplines and well-proven in financial econometrics: • Gives a precise indication of the diversity of factors • Anticipates structural changes in the markets The Intertek Group

  9. As factors change in time and new factors are called for,… • We need the tools to compare different factor models, • To understand if their differences are genuine or only apparent, • And to capture structural breaks in the market The Intertek Group

  10. Comparing different factor models: Procrustes analysis and matrix perturbation theory • Procrustes analysis and matrix perturbation theory are well known techniques in numerical analysis and pattern recognition • They compare factor models and discriminate small perturbations from irreducible differences, • And allow to represent any factor model as a perturbation of principal components analysis Benefits: • Capture structural breaks • Gauge the contribution of noisy small principal components which could be advantageously eliminated The Intertek Group

  11. Dealing with sample constraints: Random matrix theory • Financial samples allow to estimate only simple models • Hence the need to optimally choose factors so as to maximize the information captured from the data • Random matrix theory offers a solution to this problem: • It allows to separate noisy correlations / autocorrelations from meaningful correlations / autocorrelations The Intertek Group

  12. Choosing optimal factors • Random matrix theory, coupled with the analysis of the signal-to-noise ratio • Allows to determine the optimal number of factors, • And analyze just which of the factors have forecasting power, • In order to choose optimal factor models for forecasting tasks The Intertek Group

  13. Forecasting returns with factor models: Dynamic factor analysis • After analyzing just which factors have forecasting power, • We need to forecast factors and returns • This is the domain of dynamic factor analysis • Which can be applied to returns and to prices, albeit in a different form The Intertek Group

  14. Identifying sectors based on true similarities: Clustering • Classical sectors are useful but today’s markets present a structure of correlations only partially reflected in classical sectors • Clustering analysis allows to create sectors based on true similarities between firms, • Thus offering a strong basis for sector factor models The Intertek Group

  15. The Intertek Group Services The Intertek Group works in the area of advanced financial modeling, where clients can develop their own specific competitive edge. Services include: • Consulting • Training • Needs analysis, including, where appropriate, internal / external field research • Customization of models developed as a result of Intertek’s own research effort The Intertek Group

  16. Intertek Training • Cost-effective solution to develop the skills of your equity teams, be they portfolio managers or quantitative analysts • Can be used to update fundamental concepts in finance theory or financial modeling or to facilitate the deployment of new modeling methodologies • Are designed to meet the Client’s needs • Can be tailored to the level required for portfolio managers, quantitative analysts, or model developers, be it introductory or advanced • Is provided at the location and date (including weekends) that suit the Client needs The Intertek Group

  17. Training Modules • A menu of courses that can be tailored to the Client’s requirements • And come with all the software used to illustrate the concepts and techniques • Participants learn the methodology and its application to practical cases, and acquire the basic software to apply the techniques • Additional software development can be implemented to a Client’s specifications • Examples of training modules follow The Intertek Group

  18. Training Module 1: Random Matrix Theory This Master Class teaches Random Matrix Theory, a tool that allows to determine the optimal number of factors and to discriminate between factors with explanatory power and factors with predictive power. Participants will learn the basics of Random Matrix Theory and will acquire a working knowledge of how to apply the theory to enhance the forecasting power of their factor models. In particular, participants will learn how to: • Evaluate the real forecastability of returns in each market state • Separate autocorrelations and cross autocorrelations from noise • Assess the number of meaningful factors of returns • Separate factors with the highest forecasting power from factors that primarily identify risk exposures • Measure the efficiency of factors in terms of the signal-to-noise ratio The Intertek Group

  19. Training Module 2: Clustering This Master Class teaches the Clustering of time series, a technique that allows to group together stocks in function of true similarities between firms, to identify predictors, and to identify critical risk conditions. Participants will learn the basics of clustering techniques and how to apply clustering to build factor models based on true correlations. In particular, participants will learn to: • Build different measures of similarity between time series and, in particular, correlation similarity • Apply similarities to create clusters of time series • Apply different types of clustering criteria including hierarchical and k-means • Build factor models based on correlation clustering and Euclidean clustering The Intertek Group

  20. Training Module 3: Applying the Noise-to-Signal Ratio to Factor Models This Master Class addresses the problem of determining unique proprietary factors that have good explanatory power. It teaches methodologies based on the signal-to-noise ratio, a vital parameter for assessing the efficiency of factors and explaining their uniqueness. Participants will learn how to perform the signal-to-noise ratio and other vital diagnostics on their factor models. In particular, participants will learn to: • Assess the “distance” of their factors from other factors and from principal components • Gauge the uniqueness of their proprietary factors • Measure the efficiency of their factors in terms of their ability to extract correlation information from the market • Compute the maximum signal-to-noise ratio of the market The Intertek Group

  21. Training Module 4: Dynamic Factor Models Factor models can be used in forecasting returns provided that we can forecast factors or regress returns on lagged factors. Dynamic factor analysis looks at factor models specifically from the point of view of forecasting. Participants will learn the basic theory of dynamic factor models and their estimation. In particular, participants will learn to: • Build dynamic factor models of stationary variables (returns) • Build dynamic factor models of integrated variables (prices) • Determine the optimal number of factors, • Determine how to estimate factors • Forecast with dynamic factor models The Intertek Group

  22. The Intertek Group: Profile The Intertek Group provides business research, consulting, and training on modeling techniques and advanced IT technologies in the financial services sector. • The Intertek Group consulting services advises on advanced econometric techniques, model design, and implementation strategies. • The Intertek Group traininghelps in introducing new modeling methodologies with courses, in-house training, and workshops. • The Intertek Group custom research provides clients with independent, incisive analysis on technology, products and vendors. The Intertek Group

  23. Profile of Founding Partners Sergio M. Focardi is a founding partner of The Intertek Group, where he is a consultant and trains on financial modeling. Sergio is on the Editorial Board of the Journal of Portfolio Management and has (co-) authored numerous articles and books, including the CFA Institute’s monographs Challenges in Quantitative Equity Management (2008) and Trends in Quantitative Finance (2006) as well as the award-winning books Financial Modeling of the Equity Market: CAPM to Cointegration and The Mathematics of Financial Modeling and Investment Management. Most recently, Sergio co-authored Financial Econometrics – From Basics to Advanced Modeling Techniques and Robust Portfolio Optimization and Management. His research interests include the econometrics of large equity portfolios and modeling the interaction between multiple heterogeneous agents. Sergio holds a degree in Electronic Engineering from the University of Genoa. Caroline Jonas is a founding partner of The Intertek Group. She heads client research projects and is responsible for Intertek field research. Intertek projects managed by Caroline include studies on equity portfolio modeling, ALM modeling at defined-benefit pension funds, quantitative methods in asset management and unstructured data in investment management. She researched and is a co-author of the CFA Institute's survey-based monograph Challenges in Quantitative Equity Management (2008). The Intertek Group

  24. Selected books / Monographs Challenges in Quantitative Equity Management, Frank J. Fabozzi, Sergio M. Focardi, and Caroline Jonas (CFA Institute Research Foundation, May 2008). Available from www.cfapubs.org/toc/rf/2008/2008/2. This monograph has been the subject of stories published in, among others, the Wall Street Journal, Barrons, Financial Times, Financial News, Investment & Pensions Europe and The Hedge Fund Journal. Financial Econometrics: From Basics to Advanced Modeling Techniques, Svetlozar T. Rachev, Steffan Mittnik, Frank J. Fabozzi, Sergio Focardi, and T. Jasic(Wiley 2007). Robust Portfolio Optimization and Management, Frank J. Fabozzi, Peter.N. Kolm, D.A. Pachamanova, and Sergio M. Focardi (Wiley 2007). The Intertek Group

  25. Selected books / Monographs, con’t Financial Modeling of the Equity Market: From CAPM to Cointegration, Frank J. Fabozzi, Sergio Focardi, and Petter N. Kolm (Wiley 2006). Selected by Financial Engineering News as one of the top new 10 technical books in finance in 2006. Trends in Quantitative Finance, Frank J. Fabozzi, Sergio Focardi, and Petter N. Kolm (CFA Institute Research Foundation, 2006). The Mathematics of Financial Modeling and Asset Management (Wiley 2004), Sergio Focardi and Frank J. Fabozzi, Selected by RiskBook.com for “Best of 2004 Book Awards” and by Financial Engineering News as one of the top 3 books in finance in 2005. The Intertek Group

  26. Selected recent papers “ARCH/GARCH Models In Applied Financial Econometrics,” Robert F. Engle, Sergio M. Focardi, Frank J. Fabozzi, in Frank Fabozzi (Ed.), Handbook of Finance: Volume 3 (Wiley, 2008). “Black Swans and White Eagles: On Mathematics and Finance,” Sergio Focardi and Frank J. Fabozzi, forthcoming in Mathematical Methods in Operations Research. “Trends in Quantitative Equity Management: Survey Results,” Frank J. Fabozzi, Sergio M. Focardi, and Caroline Jonas, Quantitative Finance Volume 7, No. 2 (April 2007), pp. 115-122. (The results reported in this article were the subject of a feature story in CFA Institute Magazine by Susan Trammell, “Perpetual Motions: A New Study Looks at Trends in Equity Portfolio Modeling” January-February 2007, pp. 39-44. The paper is reprinted as Chapter 1 in Quantitative Fund Management edited by M. Dempster, G. Pflug, and G. Mitra, published by Taylor & Francis Group in 2008). The Intertek Group

  27. Selected recent papers, cont’d “How Do Conflicting Theories About Financial Markets Coexist?” Wesley Phoa, Sergio Focardi, Frank J. Fabozzi, Journal of Post Keynesian Economics, Vol.29, No. 4 (Summer 2007), pp. 699-701. “A Simple Framework for Time Diversification,” Frank J. Fabozzi, Sergio M. Focardi, and Petter N. Kolm, Journal of Investing (Fall 2006), pp. 8-18. “Contagion Modeling in Market and Credit Risk: Does It Add Value?” Sergio Focardi and Frank J. Fabozzi, Risk Letters Vol. 1, No. 2 (December 2005). "Implementable Quantitative Research", Frank J. Fabozzi, Sergio M. Focardi, and K.C. Ma, The Journal of Alternative Investments (Fall 2005), pp. 71-79. (Translated and reprinted as ”Überführung von quantitativem Research in implementierbare Handelsstrategien - Möglichkeiten und Grenzen der Automatisierung“, in: Michael Busack and Dieter G. Kaiser (eds.), Handbuch Alternative Investments, Band 1, Gabler Verlag, Wiesbaden.) The Intertek Group

  28. Selected recent papers, cont’d “Market Experience with Modeling for Defined-Benefit Pension Funds : Evidence from Four Countries,” Frank J. Fabozzi, Sergio M. Focardi and Caroline Jonas Journal of Pension Economics, Vol 4, No. 3 (November 2005), pp. 313-327. (Paper subject of an article in the December 20, 2004 issue of the Financial Times.) “An Autoregressive Conditional Duration Model of Credit-Risk Contagion.” Sergio M. Focardi and Frank J. Fabozzi, Journal of Risk Finance, Vol 6, No. 3 ( 2005),pp. 208-225. (Winner of the 2006 Outstanding Paper by Emerald Literati Network.) “A Methodology for Index Tracking Based on Time-Series Clustering, Sergio M. Focardi and Frank J. Fabozzi, Quantitative Finance, Vol 4, No. 4 (August 2004), pp. 417-425. The Intertek Group

  29. Selected recent papers, cont’d “A Percolation Approach to Modeling Credit Loss Distribution Under Contagion,” Sergio M. Focardi and Frank J. Fabozzi, Journal of Risk, Vol. 7, No.1 (Fall 2004), pp. 75-94. “Trends in Quantitative Asset Management in Europe,” Frank J. Fabozzi, Sergio M. Focardi, and Caroline L. Jonas, Journal of Portfolio Management, Special European Issue (Summer 2004), pp. 125-132. “Clustering Economic and Financial Time Series: Exploring the Existence of Stable Correlation Conditions,” Sergio Focardi and Frank J. Fabozzi, Finance Letters , 2004, Vol. 2, No 3, pp. 1-9. “Fat Tails, Scaling, and Stable Laws: A Critical Look at Modeling Extremal Events in Financial Phenomena?“Sergio M. Focardi and Frank J. Fabozzi, Journal of Risk Finance, Vol, 5, No.1 (Fall 2003), pp. 5-26. The Intertek Group

  30. Papers under review / in progress “Modeling Momentum and Reversals with Regime Shifting Models with Time-Varying Probabilities,” Sergio Focardi and Frank J. Fabozzi (Working Paper) “Can Returns Be Represented as an Approximate Factor Model?” Sergio Focardi and Frank J. Fabozzi, (Working Paper) “Adapting factors to market states through clustering: implications of clustering-based factor models for portfolio management” (working title). “Determining factors with forecasting power with Random Matrix Theory forecasting ability” (working title). The Intertek Group

More Related