Using Emotional Overreactions to Forecast Intraday AAPL Momentum
Summary
This paper studies whether intraday price overreactions in Apple shares can be forecast and traded as momentum signals. It combines volatility-normalized returns with emotion features derived from Twitter messages, defines overreactions relative to contemporaneous volatility and transaction costs, and frames prediction as a three-class classification task. The tested approaches include tree-based models, neural networks, and bidirectional LSTMs across several intraday sampling frequencies.
The paper reports that machine-learning models outperform benchmark overreaction rules at the shortest horizons, while classical behavioral momentum effects are stronger at intermediate frequencies, particularly around ten minutes. SHAP analysis identifies volatility and negative emotions, especially fear and sadness, as important predictors. Strategies are assessed using risk-adjusted measures and statistical tests, but the provided account gives no numerical performance results or evidence beyond AAPL, so generalization to other stocks and periods remains unclear.
Key ideas
- The study predicts three classes of overreaction using normalized returns and emotion features.
- It compares nonlinear classifiers across multiple intraday frequencies.
- Machine-learning models reportedly outperform benchmark rules at the shortest horizons.
- Classical behavioral momentum is reported to dominate at intermediate frequencies, especially around ten minutes.
- Volatility and negative emotion features are prominent in SHAP explanations.
Tags
Full text
# Overreaction as an indicator for momentum in algorithmic trading: A Case of AAPL stocks # Overreaction as an indicator for momentum in algorithmic trading: A Case of AAPL stocks 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.
Shown in full with attribution under the source's licence. Licence: abstract CC0
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.