STOCK-JEPA: Prior-Anchored Latent Revision Representation Learning in Equity Markets

2026-10-04

AI Summary

This study addresses the susceptibility of black-box models to overfitting under low signal-to-noise ratios in stock markets and the inability of linear models to capture nonlinear signals by proposing the Stock-JEPA framework. This method introduces a novel "prior-anchored incremental correction" decoupling mechanism that leverages financial priors as anchors and employs a Joint Embedding Predictive Architecture to learn predictable incremental correction representations, thereby integrating structural interpretability with nonlinear modeling capacity. Furthermore, it theoretically proves that the correction term strictly reduces prediction error. Extensive experiments on large-scale Chinese and U.S. stock markets demonstrate that the proposed framework significantly outperforms thirteen strong baselines, effectively enhancing cross-sectional ranking accuracy and portfolio performance. These results validate the efficacy of complementing economic structure with deep learning for financial forecasting.

Abstract

Learning effective representations helps characterize the structure and dynamics of equity markets from financial data with a low signal-to-noise ratio. Black-box deep models can capture complex patterns but may overfit sample noise and lack explicit economic structure. Meanwhile, classic linear financial models provide interpretable references, but their oversimplified assumptions leave non-linear signals uncaptured. To combine the strengths of these two directions, we propose Stock-JEPA, a joint-embedding predictive framework that learns predictable incremental revisions relative to a point-in-time financial prior. First, we leverage a low-complexity financial model to produce fixed statistics summarizing multi-horizon return and risk. A prior projector then maps these statistics into the target encoder's latent space as an anchor. Second, we design a context-conditioned revision predictor to estimate the future representation's predictable displacement from the anchor. Separate losses update the two branches: the anchor learns from prior statistics, while the revision captures additional predictable information from historical context. Third, we freeze all representation modules and train a downstream readout, evaluating its forecasts through cross-sectional ranking and portfolio performance. Theoretically, we prove that optimal revision reduces the prior anchor's expected squared error for the same future representation by exactly $\mathbb{E}[\|\boldsymbolΔ\|_2^2]$. This non-negative gain is the expected squared magnitude of the additional signal predictable from historical context. Experimentally, Stock-JEPA outperforms 13 strong baselines across large-scale China and U.S. equity universes on 5 key evaluation metrics. Ablation studies and representation analysis further demonstrate the value of the learned revisions for representation learning in equity markets.

Problem

• representation learning
• equity markets
• low signal-to-noise ratio
• financial prior
• non-linear signals

Innovation

• Joint-Embedding Predictive Architecture
• Prior-Anchored Representation Learning
• Latent Revision Prediction
• Context-Conditioned Predictor
• Equity Markets