Deep Neural Networks for Robust Statistical Arbitrage

Summary

The document describes a data-driven method for finding statistical arbitrage strategies that seek to remain profitable under uncertainty about the market model. It uses deep neural networks to consider many securities together, avoiding the need to first identify cointegrated pairs. This broadens the approach to higher-dimensional markets and settings where conventional pairs trading may not work well.

The method also constructs a set of plausible probability measures from observed market data to represent model ambiguity. The authors characterize the resulting approach as model-free. They report empirical investigations with profitable performance in dimensions reaching 50, including during financial crises and after pairwise cointegration relationships cease to persist. The excerpt does not provide datasets, evaluation design, cost assumptions, or detailed performance measures, so it is not enough to assess robustness in live trading or reproduce the findings.

Key ideas

• Deep neural networks are used to identify statistical arbitrage strategies under model ambiguity.
• The method can consider many securities without requiring cointegrated pairs.
• An ambiguity set of plausible probability measures is derived from observed data.
• The reported empirical investigations include crisis periods and cases where cointegration breaks down.
• The excerpt omits testing details and trading cost assumptions needed to assess practical performance.

Tags

# Detecting data-driven robust statistical arbitrage strategies with deep neural networks


# Detecting data-driven robust statistical arbitrage strategies with deep neural networks









We present an approach, based on deep neural networks, that allows identifying robust statistical arbitrage strategies in financial markets. Robust statistical arbitrage strategies refer to trading strategies that enable profitable trading under model ambiguity. The presented novel methodology allows to consider a large amount of underlying securities simultaneously and does not depend on the identification of cointegrated pairs of assets, hence it is applicable on high-dimensional financial markets or in markets where classical pairs trading approaches fail. Moreover, we provide a method to build an ambiguity set of admissible probability measures that can be derived from observed market data. Thus, the approach can be considered as being model-free and entirely data-driven. We showcase the applicability of our method by providing empirical investigations with highly profitable trading performances even in 50 dimensions, during financial crises, and when the cointegration relationship between asset pairs stops to persist.

Shown in full with attribution under the source's licence. Licence: abstract CC0

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.