When quantitative trading meets machine learning: A pilot survey

Abstract

Quantitative trading strategies are designed to look for relationships between data about an underlying security and its future price and then to generate alpha on a trading desk. Recent years have witnessed the increasing attention from both academic and corporate sectors on enhancing quantitative trading by machine learning techniques due to their excellent predictive powers, with a few successful stories from the markets further boosting optimism for this method of analysis. In this paper, we aim to conduct a comprehensive survey on the pilot study of applying machine learning for quantitative trading. We will review some earlier studies of using NNs and SVMs for stock price prediction. We will also touch some recent studies on designing online learning algorithms based on characteristics of financial time series, e.g., mean reversion of stock price. Another application of machine learning in quantitative trading is called meta-learning algorithm which considers how to assign weights to strategies. We will finally summarize the above research by pointing out promising machine learning techniques for different categories of trading strategies. We will also discuss slightly the potentials of machine learning techniques in helping generate strategies that do not only base on financial market data, like behavioral strategy, event-driven and untraditional index strategy.