Enhancing Stock Price Prediction with MACD and EMA Features Using LSTM Algorithm

This research paper proposes the application of the Moving Average Convergence Divergence (MACD) and Exponential Moving Average (EMA) indicators in combination with the Long Short-Term Memory (LSTM) algorithm for stock price prediction. This study aims to enhance the accuracy and robustness of stock market predictions by using the strengths of technical indicators and deep learning methodologies. The LSTM algorithm, known for its ability to model long-term time series data, is used in this research. In this paper, we considered various evaluation metrics such as RMSE, R2 score, accuracy, precision, recall, and F1 score. Real-world historical stock price data, i.e., RELIANCE, KOTAK BANK, INFOSYS, ASIAN PAINT, HINDUSTAN UNILEVER, HDFC BANK, VEDANTA, and ICICI BANK are utilized to validate the proposed model, ensuring practical relevance and applicability. The experimental results demonstrate that the integration of MACD and different variants of EMA (26-day EMA, 100-day EMA, and 200-day EMA) features with LSTM significantly improves stock price prediction accuracy. The results show that the accuracy is improved by using the indicators MACD and EMA, as well as basic features such as open, high, low, and close (OHLC) features. Good precision values indicate that the proposed model captures positive events effectively. Good recall values indicate that the model can predict true positive values correctly, and the F1-score is also good, which indicates that the model has a good combination for effectively capturing positive and negative instances. The RMSE is computed on a testing dataset to assess the model's performance. A lower RMSE indicates better performance. The R2 score was 0.75 < R2 < 0.95, which indicated a strong model.