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’ll discuss some of the challenges that come from applying modern ML techniques to financial markets. We will provide some background about the massive, heterogeneous, unevenly spaced, noisy, and bursty tick-by-tick datasets that many of our machine learning algorithms process. We’ll explore applications of contrastive learning, which seeks to learn feature embeddings in which similar inputs have similar feature representations, for learning feature representations of related assets for downstream tasks. Next, we’ll talk about applications of meta-learning for financial timeseries models, such as quickly adapting to new financial products. Finally, we’ll discuss some of the issues that need to be overcome to apply state-of-the-art NLP techniques to problems in forecasting.