Stock forecasting improved by multi-task learning with liquidity signals
Hierarchical Multi-Task Learning with Liquidity-Aware Signals for Stock Forecasting
Computational Engineering, Finance, and Science
Summary
Stock prices are hard to predict because markets are random and complex. The authors created a new method called LiMT that looks at relationships between stocks and how each stock changes over time, while also considering how easily stocks can be traded (liquidity). Their method predicts price changes, volatility, and trading volume together, and then turns these predictions into actionable investment strategies that account for costs and liquidity. Tests show this approach can lead to better investment returns and risk levels.
What this means in practice
- •For quantitative trading teams: Use multi-task predictions of price, volatility, and volume to create optimized portfolios under liquidity and transaction cost constraints.
- •For financial software developers: Develop advanced stock forecasting tools that improve prediction accuracy by modeling cross-stock relationships and liquidity-sensitive signals.$Commercial implications: This paper enables building enhanced forecasting software that delivers better investment decision support for trading firms.
Authors
Hengyi Yang, Sida Lin, Yiyan Qi, Yankai Chen, Haohan Zhang, Xianhua Peng, Jian Guo
Abstract
Stock price forecasting is a long-standing challenge in computational finance, driven by the inherent randomness of markets and complex temporal patterns. While recent deep-learning models have raised forecasting accuracy by jointly modeling inter-stock and temporal price dynamics, they conflate inter-stock relationships with intra-stock temporal dependencies and focus solely on the univariate objective of price movement. To address these limitations, we propose LiMT, a Hierarchical Multi-Task Learning framework that integrates liquidity-aware signals for stock price forecasting. LiMT employs a Market Regime Encoder (MRE) module that first extracts contemporaneous cross-stock dependencies, then models each stock's temporal dynamics, yielding a unified latent state. Building on this latent state, we introduce a Liquidity-Driven Learning (LDL) module, a mixture-of-experts architecture that features cross-task gating mechanisms to jointly predict price movement, volatility, and trading volume. We further design an Adaptive Portfolio Optimization (APO) mechanism that converts multi-task forecasts into executable portfolio weights under transaction-cost and liquidity constraints. Extensive experiments on the CSI300 and CSI500 benchmarks show that LiMT performs best among strong neural and tree-based baselines across the reported metrics. In realistic CSI300 backtests, APO improves annualized return from 3.99% to 10.01% and Sharpe ratio from 1.22 to 1.86 over equal weighting, showing that the multi-task forecasts translate into deployable portfolio gains.