Overreaction as an indicator for momentum in algorithmic trading: A Case of AAPL stocks

2026-02-21

AI Summary

This study investigates whether short-term market overshooting can be systematically predicted and leveraged as a momentum trading signal. Using high-frequency Apple stock data, the authors model overshooting as a three-class classification problem by integrating Twitter sentiment features extracted via Transformer with volatility-standardized returns. They employ nonlinear models—including XGBoost, Random Forest, deep neural networks, and bidirectional LSTM—to predict overshooting events, enhancing interpretability through SHAP analysis. The findings reveal that negative emotions, particularly fear and sadness, significantly drive intraday momentum formation. At a 10-minute frequency, the predictive models substantially outperform benchmark rules, and the resulting trading strategy delivers strong risk-adjusted performance, confirming that emotion-driven market overshooting exhibits a predictable structure.

Technology Category

Data Mining & Knowledge Management: Anomaly/Outlier DetectionMachine Learning: Time-Series/Data StreamsCognitive Modeling & Cognitive Systems: Affective Computing

Application Category

Economics, Online Markets and Human Computation: The sharing economyWeb Mining and Content Analysis: Sentiment analysis and opinion miningSocial Networks and Social Media: Influence propagation, information diffusion, and the prediction on networks

Abstract

This paper investigates whether short-term market overreactions can be systematically predicted and monetized as momentum signals using high-frequency emotional information and modern machine learning methods. Focusing on Apple Inc. (AAPL), we construct a comprehensive intraday dataset that combines volatility normalized returns with transformer-based emotion features extracted from Twitter messages. Overreactions are defined as extreme return realizations relative to contemporaneous volatility and transaction costs and are modeled as a three-class prediction problem. We evaluate the performance of several nonlinear classifiers, including XGBoost, Random Forests, Deep Neural Networks, and Bidirectional LSTMs, across multiple intraday frequencies (1, 5, 10, and 15 minute data). Model outputs are translated into trading strategies and assessed using risk-adjusted performance measures and formal statistical tests. The results show that machine learning models significantly outperform benchmark overreaction rules at ultra short horizons, while classical behavioral momentum effects dominate at intermediate frequencies, particularly around 10 minutes. Explainability analysis based on SHAP reveals that volatility and negative emotions, especially fear and sadness, play a central role in driving predicted overreactions. Overall, the findings demonstrate that emotion-driven overreactions contain a predictable structure that can be exploited by machine learning models, offering new insights into the behavioral origins of intraday momentum and the interaction between sentiment, volatility, and algorithmic trading.

Problem

• overreaction
• momentum
• algorithmic trading
• market sentiment
• intraday prediction

Innovation

• overreaction
• sentiment analysis
• machine learning
• intraday momentum
• behavioral finance