ACT: Anti-Crosstalk Learning for Cross-Sectional Stock Ranking via Temporal Disentanglement and Structural Purification

TL;DR

This paper introduces ACT, a novel framework that disentangles temporal components and purifies structural relations to improve cross-sectional stock ranking accuracy.

Contribution

It proposes a new method combining temporal disentanglement and structural purification to address crosstalk in graph-based stock ranking models.

Findings

Achieves state-of-the-art ranking accuracy on CSI300 and CSI500 datasets.

Improves portfolio performance with up to 74.25% gains on CSI300.

Effectively decouples non-transferable local patterns and relation-specific signals.

Abstract

Cross-sectional stock ranking is a fundamental task in quantitative investment, relying on both temporal modeling of individual stocks and the capture of inter-stock dependencies. While existing deep learning models leverage graph-based approaches to enhance ranking accuracy by propagating information over relational graphs, they suffer from a key challenge: crosstalk, namely unintended information interference across predictive factors. We identify two forms of crosstalk: temporal-scale crosstalk, where trends, fluctuations, and shocks are entangled in a shared representation and non-transferable local patterns contaminate cross-stock learning; and structural crosstalk, where heterogeneous relations are indiscriminately fused and relation-specific predictive signals are obscured. To address both issues, we propose the Anti-CrossTalk (ACT) framework for cross-sectional stock ranking via temporal disentanglement and structural purification. Specifically, ACT first decomposes each stock sequence into trend, fluctuation, and shock components, then extracts component-specific information through dedicated branches, which effectively decouples non-transferable local patterns. ACT further introduces a Progressive Structural Purification Encoder to sequentially purify structural crosstalk on the trend component after mitigating temporal-scale crosstalk. An adaptive fusion module finally integrates all branch representations for ranking. Experiments on CSI300 and CSI500 demonstrate that ACT achieves state-of-the-art ranking accuracy and superior portfolio performance, with improvements of up to 74.25% on the CSI300 dataset.