FURI | Spring 2025

Comparing LSTMs with and without Reinforcement Learning for Stock Price Prediction and Trading Decisions

Long Short-Term Memory (LSTM) neural networks are widely used for stock price prediction, and the addition of Reinforcement Learning (RL) is often assumed to enhance performance, but there is little research quantifying its actual impact. This project compares LSTM-only models with LSTM+RL models, evaluating their predictive accuracy and trading performance using historical S&P 500 data. By measuring key metrics such as Mean Absolute Error and simulated trading returns, this study will provide quantitative insight into whether RL meaningfully improves LSTM-based trading indicators, guiding more informed decision-making in algorithmic trading and quantitative finance.