Learning Explainable Stock Predictions with Tweets Using Mixture of Experts

2025-07-28

AI Summary

To address the challenge of long-horizon, multi-granularity stock price forecasting using heterogeneous multimodal data—namely historical price series and unstructured textual data (e.g., news and social media)—this paper proposes FTS-Text-MoE, a novel forecasting framework. First, it extracts salient textual summaries and constructs point-wise embeddings, which are jointly fed with time-series inputs. Second, it introduces a Mixture-of-Experts (MoE) Transformer decoder to enhance modeling capacity while reducing computational overhead. Third, it incorporates a multi-resolution prediction head to support flexible, scale-aware trend forecasting. Extensive experiments on real-world financial datasets demonstrate that FTS-Text-MoE significantly improves forecasting accuracy over state-of-the-art baselines. Moreover, it achieves superior risk-adjusted returns—evidenced by higher investment profitability and Sharpe ratio—while maintaining computational efficiency, interpretability, and practical deployability.

Technology Category

• Machine Learning: Mixture of Experts (MoE)
• Natural Language Processing: Language Grounding & Multi-modal NLP
• Data Mining & Knowledge Management: Mining of Visual, Multimedia & Multimodal Data

Application Category

• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• Web Mining and Content Analysis: Large pretrained models with web data
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Stock price movements are influenced by many factors, and alongside historical price data, tex-tual information is a key source. Public news and social media offer valuable insights into market sentiment and emerging events. These sources are fast-paced, diverse, and significantly impact future stock trends. Recently, LLMs have enhanced financial analysis, but prompt-based methods still have limitations, such as input length restrictions and difficulties in predicting sequences of varying lengths. Additionally, most models rely on dense computational layers, which are resource-intensive. To address these challenges, we propose the FTS- Text-MoE model, which combines numerical data with key summaries from news and tweets using point embeddings, boosting prediction accuracy through the integration of factual textual data. The model uses a Mixture of Experts (MoE) Transformer decoder to process both data types. By activating only a subset of model parameters, it reduces computational costs. Furthermore, the model features multi-resolution prediction heads, enabling flexible forecasting of financial time series at different scales. Experimental results show that FTS-Text-MoE outperforms baseline methods in terms of investment returns and Sharpe ratio, demonstrating its superior accuracy and ability to predict future market trends.

Problem

• Predict stock prices using tweets and news with limited input length
• Reduce computational costs in financial prediction models
• Forecast financial time series at multiple scales flexibly

Innovation

• Combines numerical data with textual summaries using embeddings
• Uses Mixture of Experts Transformer for efficient processing
• Features multi-resolution heads for flexible time series forecasting