A Neuro-Fuzzy System for Interpretable Long-Term Stock Market Forecasting

2025-10-01

AI Summary

Addressing the challenge of balancing accuracy and interpretability in long-term multivariate time series forecasting, this paper proposes NF-LSTM-Attn, a neuro-fuzzy system integrating LSTM, temporal multi-head self-attention, and fuzzy inference. Methodologically, multi-head self-attention is innovatively embedded within the LSTM architecture to enhance modeling of complex temporal dependencies, while a dedicated fuzzy rule module explicitly encodes semantic relationships between key input features and predictions—thereby rendering the model’s internal decision-making process interpretable. Experimental evaluation on S&P 500 index data demonstrates that NF-LSTM-Attn achieves forecasting accuracy comparable to ARIMA and standard LSTM. Crucially, it additionally provides human-readable fuzzy rules and attention-weight visualizations, significantly improving model transparency, trustworthiness, and practical utility in financial applications.

Technology Category

• Machine Learning: Neuro-Symbolic Learning
• Natural Language Processing: Interpretability, Analysis, and Evaluation of NLP Models
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• User Modeling, Personalization and Recommendation: Explainable and interpretable methods for personalization

Abstract

In the complex landscape of multivariate time series forecasting, achieving both accuracy and interpretability remains a significant challenge. This paper introduces the Fuzzy Transformer (Fuzzformer), a novel recurrent neural network architecture combined with multi-head self-attention and fuzzy inference systems to analyze multivariate stock market data and conduct long-term time series forecasting. The method leverages LSTM networks and temporal attention to condense multivariate data into interpretable features suitable for fuzzy inference systems. The resulting architecture offers comparable forecasting performance to conventional models such as ARIMA and LSTM while providing meaningful information flow within the network. The method was examined on the real world stock market index S&P500. Initial results show potential for interpretable forecasting and identify current performance tradeoffs, suggesting practical application in understanding and forecasting stock market behavior.

Problem

• Achieving accurate and interpretable multivariate time series forecasting
• Developing a neuro-fuzzy system for long-term stock market prediction
• Combining neural networks with fuzzy inference for interpretable features

Innovation

• Combines fuzzy inference with neural networks
• Uses LSTM and attention for feature extraction
• Provides interpretable stock market forecasting