Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

CSPO: Cross-Market Synergistic Stock Price Movement Forecasting with Pseudo-volatility Optimization

Sida Lin, Yankai Chen, Yiyan Qi, Chenhao Ma, Bokai Cao, Yifei Zhang, Xue Liu, Jian Guo

TL;DR

CSPO tackles stock price movement forecasting under two key challenges: exogenous cross-market influences and volatility heterogeneity. It introduces a Bi-level Dense Pricing Transformer (BDP-Former) that fuses futures-market signals with stock data, plus a pseudo-volatility (PV)–guided loss to reflect forecast confidence and stock-specific risk. Extensive proprietary backtesting and public benchmarking show CSPO achieving superior information metrics and financially favorable outcomes, with ablations confirming the value of futures fusion and volatility-aware optimization. The work offers practical impact for alpha-factor generation and trading strategies, and points to future enhancements through large-language-model integration, graph data management, and continual learning for streaming markets.

Abstract

The stock market, as a cornerstone of the financial markets, places forecasting stock price movements at the forefront of challenges in quantitative finance. Emerging learning-based approaches have made significant progress in capturing the intricate and ever-evolving data patterns of modern markets. With the rapid expansion of the stock market, it presents two characteristics, i.e., stock exogeneity and volatility heterogeneity, that heighten the complexity of price forecasting. Specifically, while stock exogeneity reflects the influence of external market factors on price movements, volatility heterogeneity showcases the varying difficulty in movement forecasting against price fluctuations. In this work, we introduce the framework of Cross-market Synergy with Pseudo-volatility Optimization (CSPO). Specifically, CSPO implements an effective deep neural architecture to leverage external futures knowledge. This enriches stock embeddings with cross-market insights and thus enhances the CSPO's predictive capability. Furthermore, CSPO incorporates pseudo-volatility to model stock-specific forecasting confidence, enabling a dynamic adaptation of its optimization process to improve accuracy and robustness. Our extensive experiments, encompassing industrial evaluation and public benchmarking, highlight CSPO's superior performance over existing methods and effectiveness of all proposed modules contained therein.

CSPO: Cross-Market Synergistic Stock Price Movement Forecasting with Pseudo-volatility Optimization

TL;DR

CSPO tackles stock price movement forecasting under two key challenges: exogenous cross-market influences and volatility heterogeneity. It introduces a Bi-level Dense Pricing Transformer (BDP-Former) that fuses futures-market signals with stock data, plus a pseudo-volatility (PV)–guided loss to reflect forecast confidence and stock-specific risk. Extensive proprietary backtesting and public benchmarking show CSPO achieving superior information metrics and financially favorable outcomes, with ablations confirming the value of futures fusion and volatility-aware optimization. The work offers practical impact for alpha-factor generation and trading strategies, and points to future enhancements through large-language-model integration, graph data management, and continual learning for streaming markets.

Abstract

The stock market, as a cornerstone of the financial markets, places forecasting stock price movements at the forefront of challenges in quantitative finance. Emerging learning-based approaches have made significant progress in capturing the intricate and ever-evolving data patterns of modern markets. With the rapid expansion of the stock market, it presents two characteristics, i.e., stock exogeneity and volatility heterogeneity, that heighten the complexity of price forecasting. Specifically, while stock exogeneity reflects the influence of external market factors on price movements, volatility heterogeneity showcases the varying difficulty in movement forecasting against price fluctuations. In this work, we introduce the framework of Cross-market Synergy with Pseudo-volatility Optimization (CSPO). Specifically, CSPO implements an effective deep neural architecture to leverage external futures knowledge. This enriches stock embeddings with cross-market insights and thus enhances the CSPO's predictive capability. Furthermore, CSPO incorporates pseudo-volatility to model stock-specific forecasting confidence, enabling a dynamic adaptation of its optimization process to improve accuracy and robustness. Our extensive experiments, encompassing industrial evaluation and public benchmarking, highlight CSPO's superior performance over existing methods and effectiveness of all proposed modules contained therein.

Paper Structure

This paper contains 33 sections, 22 equations, 8 figures, 7 tables.

Table of Contents

  1. Introduction
  2. Problem Formulation
  3. CSPO Framework
  4. Overview
  5. Market Time Series Encoding
  6. Bi-level Dense Pricing Transformer
  7. Cross-market Embedding Layer
  8. Price Movement Forecasting Layer
  9. Pseudo-volatility Optimization
  10. Pseudo-volatility Estimation.
  11. Volatility-aware Regression Optimization
  12. Experiments
  13. Proprietary Backtesting Evaluation (RQ1)
  14. Overview.
  15. Evaluation Data.
  16. ...and 18 more sections

Figures (8)

  • Figure 1: (a) Stock price movement forecasting (SPMF). (b) Portfolio yield comparison between baselines, i.e., CSI300 Index and SSE Composite, and our method to use CSPO results as alpha-factors for hedged and spot return.
  • Figure 2: An illustration of our framework overview (best view in color). In CME-Layer, the notations, e.g., $\boldsymbol{W}^Q$, generalize to both cases of commodity futures and financial futures, e.g., $\boldsymbol{W}_c^Q$ and $\boldsymbol{W}_e^Q$.
  • Figure 3: Hedged return curve.
  • Figure 4: Spot return curve.
  • Figure 6: CSI300 yield curve.
  • ...and 3 more figures

Theorems & Definitions (2)

  • Definition 1: Stock Market Time Series
  • Definition 2: Futures Market Time Series