Portfolio Construction from Predicted Distributions of Residual Factors

Summary

This study proposes a stock portfolio method based on predicting the distribution of residual factors, which capture information beyond common market exposures and can help hedge those exposures. It describes two main components: a computationally efficient way to extract residual information and a neural network architecture designed to incorporate amplitude and time-scale invariance. The stated motivation is to make machine learning more sample-efficient and robust in non-stationary financial markets.

The authors evaluate the approach on U.S. and Japanese stock market data and report that ablation experiments find contributions from each of the proposed techniques to trading-strategy performance. The document does not give portfolio construction details, the prediction target horizon, performance measures, or comparisons with specific benchmarks. Its evidence therefore supports the described components within the reported experiments, but the brief account does not establish how the results generalize across markets or trading costs.

Key ideas

• The portfolio method predicts distributions of residual factors to address risk beyond common market exposures.
• A computationally efficient extraction method produces residual information for prediction models.
• The neural network incorporates amplitude and time-scale invariance as financial inductive biases.
• Experiments use U.S. and Japanese stock market data.
• Ablation results indicate that each proposed technique contributes to strategy performance.

Tags

# Deep Portfolio Optimization via Distributional Prediction of Residual Factors


# Deep Portfolio Optimization via Distributional Prediction of Residual Factors









Recent developments in deep learning techniques have motivated intensive research in machine learning-aided stock trading strategies. However, since the financial market has a highly non-stationary nature hindering the application of typical data-hungry machine learning methods, leveraging financial inductive biases is important to ensure better sample efficiency and robustness. In this study, we propose a novel method of constructing a portfolio based on predicting the distribution of a financial quantity called residual factors, which is known to be generally useful for hedging the risk exposure to common market factors. The key technical ingredients are twofold. First, we introduce a computationally efficient extraction method for the residual information, which can be easily combined with various prediction algorithms. Second, we propose a novel neural network architecture that allows us to incorporate widely acknowledged financial inductive biases such as amplitude invariance and time-scale invariance. We demonstrate the efficacy of our method on U.S. and Japanese stock market data. Through ablation experiments, we also verify that each individual technique contributes to improving the performance of trading strategies. We anticipate our techniques may have wide applications in various financial problems.

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.