Jeremy Evnine, CEO and President of Evnine & Associates
Reflections on Stat-Arb from a Machine Learning Perspective
•I have been involved in Statistical Arbitrage since 1992. I ran a firm whose principle activity was stat-arb until 2011. Since then, there has been an explosion of (a) Machine Learning methods, and (b) computing power.
•This has caused me to reevaluate everything I thought I knew about stat-arb, and to consider how these advances could be exploited in the context of stat-arb.
•The stat-arb problem differs in several important respects from many of the prediction problems to which ML has been applied, but with some creativity ML methods can be adapted to improve stat-arb predictions.
Jeremy is currently CEO of Evnine & Associates, Inc., a consulting and Investment Advisory firm engaged in quantitative strategies since 1992. From 1991 to 2003, Jeremy was also a partner in Iris Financial Engineering and Systems, a financial software firm specializing in providing high-end trading and risk systems to top-tier investment banks. He sold his interest in Iris in 2003.
From 1984-1990, Jeremy was SVP in charge of research at WFIA (subsequently Barclays Global Investors, now BlackRock).
From 1980-1984, Jeremy was a consultant at Barra, where he developed the firm’s option products.
Jeremy also serves as a Trustee of the City of San José Police and Fire Department Retirement Plan.
Jeremy earned his B.Sc. in Mathematics at Manchester University in England, his M.Sc. in Pure Mathematics at the Hebrew University of Jerusalem, and his Ph.D. in Operations Research and Finance at U.C. Berkeley. He has taught
courses in finance at U.C. Berkeley, published articles in the financial literature on option pricing and tactical asset allocation, and lectured in the United States and abroad.