MIGA: Mixture-of-Experts with Group Aggregation for Stock Market Prediction
Zhaojian Yu, Yinghao Wu, Genesis Wang, Heming Weng
2024-10-07
Summary
This paper introduces MIGA, a new framework that uses a mixture of expert models to improve stock market predictions by dynamically switching between specialized models for different types of stocks.
What's the problem?
Predicting stock market movements is very challenging due to the unpredictable nature of stocks and the noise in the data. Traditional machine learning methods often use a single model for all stocks, which can lead to inaccurate predictions because different stocks behave differently and have unique trends. This one-size-fits-all approach fails to capture the specific characteristics of various stocks.
What's the solution?
To solve this problem, the authors developed MIGA, which stands for Mixture-of-Experts with Group Aggregation. This framework allows the model to switch between different 'expert' models that specialize in various stock styles. MIGA also includes a unique inner group attention mechanism that helps these experts share information with each other, improving their overall performance. The authors tested MIGA on three major Chinese stock indices and found that it significantly outperformed traditional models, achieving a 24% excess annual return on one of the benchmarks.
Why it matters?
This research is important because it demonstrates how using specialized models can enhance stock market predictions, leading to better investment strategies. By improving the accuracy of stock predictions, MIGA could help investors make more informed decisions, ultimately contributing to more effective financial markets.
Abstract
Stock market prediction has remained an extremely challenging problem for many decades owing to its inherent high volatility and low information noisy ratio. Existing solutions based on machine learning or deep learning demonstrate superior performance by employing a single model trained on the entire stock dataset to generate predictions across all types of stocks. However, due to the significant variations in stock styles and market trends, a single end-to-end model struggles to fully capture the differences in these stylized stock features, leading to relatively inaccurate predictions for all types of stocks. In this paper, we present MIGA, a novel Mixture of Expert with Group Aggregation framework designed to generate specialized predictions for stocks with different styles by dynamically switching between distinct style experts. To promote collaboration among different experts in MIGA, we propose a novel inner group attention architecture, enabling experts within the same group to share information and thereby enhancing the overall performance of all experts. As a result, MIGA significantly outperforms other end-to-end models on three Chinese Stock Index benchmarks including CSI300, CSI500, and CSI1000. Notably, MIGA-Conv reaches 24 % excess annual return on CSI300 benchmark, surpassing the previous state-of-the-art model by 8% absolute. Furthermore, we conduct a comprehensive analysis of mixture of experts for stock market prediction, providing valuable insights for future research.