Dependence-Informed Sparse Neural Architecture for Stock Return Prediction

2026-08-14

AI Summary

This study addresses the lack of financial interpretability in neural network-based stock prediction by proposing a sparse architecture grounded in feature dependency graphs. By estimating dependencies via MFCF and mapping them onto HNN topology, the method aligns network structure with financial logic while controlling complexity through clique size. Experimental results demonstrate that the model achieves predictive accuracy comparable to benchmarks while delivering superior ranking performance. Notably, it reduces parameter count by approximately 80-fold and significantly enhances both sparsity and feature grouping. This work effectively balances deep learning performance with financial interpretability, offering a lightweight, structured paradigm for asset pricing that reconciles computational efficiency with domain-specific transparency.

Abstract

Using neural networks for stock return prediction typically requires choices about depth and hidden-layer width that are difficult to connect to financial interpretation. We study an alternative: estimate dependence among firm characteristics with a Maximally Filtered Clique Forest (MFCF), then map its clique structure to a Homological Neural Network (HNN). The MFCF maximum clique size K is the only parameter controlling architectural complexity, and it has a clear graphical meaning: it bounds the number of characteristics in each maximal clique and hence the highest interaction order the network can represent. The filtered graph then fixes the neural network's depth, layer widths, and sparse connections before training, in place of a separately chosen depth and width sequence. We apply two HNN variants to annual out-of-sample forecasts of U.S. stock excess returns from 1987 to 2016 using 94 firm characteristics. The HNN models match a three-hidden-layer benchmark on pooled predictive accuracy, rank the cross-section more accurately, and use roughly 80 times fewer parameters than a fully connected network with the same induced layer widths. Two structural ablations indicate that both the sparse connectivity and the estimated grouping of characteristics contribute to the ranking advantage, and both effects remain significant after correcting for multiple testing. These findings show that HNNs offer a practical and interpretable way to incorporate estimated dependence among firm characteristics into neural architecture design.

Problem

• Stock Return Prediction
• Neural Architecture Design
• Financial Interpretability
• Firm Characteristics Dependence

Innovation

• Homological Neural Network
• Maximally Filtered Clique Forest
• Sparse Neural Architecture
• Interpretability
• Stock Return Prediction