Deep Learning Statistical Arbitrage in Market Capitalization Rank Space

Summary

This work proposes representing equities by their rank in market capitalization rather than by company identity. The motivation is that conventional name-based representations can be noisy and volatile, making patterns difficult for deep neural networks to learn. In rank space, stocks are organized by relative capitalization position, creating a transformed view of market behavior for statistical arbitrage research.

The authors report that deep neural networks perform better for statistical arbitrage in rank space than in name space. They attribute the improvement to more robust market representations and stronger mean-reverting behavior in residual returns, which can make learning more effective. The description does not specify the universe, sample period, trading rules, costs, or performance measures, and it provides no numeric results. The claimed advantage should therefore be understood as a reported comparative finding, not evidence that rank-based models will outperform across markets or under live execution.

Key ideas

• The method indexes stocks by capitalization rank instead of company name.
• The transformation aims to reduce noise and improve the structure of market representations.
• Deep neural networks are reported to perform better for statistical arbitrage in rank space.
• The authors link the gain to stronger mean reversion in residual returns.
• The supplied description omits trading costs, sample details, and numerical performance evidence.

Tags

# Statistical Arbitrage in Rank Space


# Statistical Arbitrage in Rank Space









Equity market dynamics are conventionally investigated in name space, where stocks are indexed by company names. However, this perspective often suffers from high volatility and a low signal-to-noise ratio, which poses challenges for effective learning by deep neural networks (DNNs). In contrast, by indexing stocks by their ranks in capitalization, we gain a distinct and more structured view of market behavior in rank space. In this work, we demonstrate that DNNs achieve superior performance in statistical arbitrage when operating in rank space compared to name space. This performance gain is driven by more robust market representations and enhanced mean-reverting properties of residual returns in rank space, which facilitate more efficient learning. Our findings highlight the critical role of domain-informed data transformation in improving deep learning performance in noisy financial environments.

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.