Stock market prediction using the LSTM algorithm in association with the Relative Strength Index (RSI) and Exponential Moving Average (EMA) indicators.

Abstract

Abstract Because of the unpredictable nature of the financial market, stock prediction is very difficult. To invest investors' hard-earned money in the financial market, we require additional information. Traditional models like linear regression and Support Vector Regression (SVR) are used to predict stock prices, but they do not have much accuracy. Recurrent Neural Network (RNN) is having "vanishing gradient" issues. In this study, we explain the technique of combining the Long Short-Term Memory (LSTM) machine learning algorithm with leading indicators like the Relative Strength Index (RSI) and the Exponential Moving Average (EMA), i.e., the slow EMA, medium EMA, and fast EMA. For this study, we have selected seven different stocks from the National Stock Exchange (NSE), and the dataset period is from January 1, 2012, to December 31, 2022. When we add extra features like RSI, 50-day EMA, 100-day EMA, and 150-day EMA to traditional ones like open, high, low, close, and volume, we get better results than when we only use traditional ones like open, high, low, close, and volume. When the indicators are added along with the traditional features, the Mean Absolute Percentage Error (MAPE) goes down, the R2 score (coefficient of determination) goes up, and the model does better than the conventional model. This study and analysis helps to improve intraday trading by predicting the value and trend of certain stocks.