NIPS 2018 Expo Talk
Dec. 6, 2020
Inside the Black Box: Machine Learning and Optimization at HRT
Sponsor: Hudson River Trading
Iain Dunning (HRT), Vlad Kontsevoi (HRT), Minyu Peng (HRT), Xiaoqi Zhu (HRT)
Hudson River Trading (HRT) is a quantitative automated trading company that trades hundreds of millions of shares each day broken up into over a million trades and spread across thousands of symbols. It trades on about 100 markets worldwide, and accounts for over 5% of US equities volume. To provide price discovery and market making services for public markets, HRT employs state-of-the-art techniques from machine learning and optimization to understand market data.
In this talk we will describe the nature of some of our problems, and in particular how they differ from problems commonly encountered in the literature. First, we’ll explain and visualize the massive, heterogeneous, unevenly spaced, noisy, and bursty tick-by-tick datasets that many of our machine learning algorithms process. Next, we’ll discuss the application of reinforcement learning to making better trading decisions, and why many common approaches fail. Lastly, drawing upon ideas from convex optimization and insights into market structure, we will demonstrate how we can construct portfolios based on a combination of predictive signals over different time horizons in the face of predictable patterns in risk and liquidity.