Stock Price Prediction using LSTM and TLBO

Abstract

The stock market's inherent complexity and volatility make accurate predictions difficult. Complex patterns in unstructured data may be simulated and analyzed with the help of Deep Learning. It is possible to employ deep learning models for a wide variety of purposes, including but not limited to image recognition, audio recognition, and natural language processing (NLP). As a result, everyone with a stake in the market is always trying to make educated guesses about what the stock price will do next. We used many deep learning techniques, including Long Short-Term Memory, to model our problem (LSTM). To successfully predict the closing prices of three distinct markets, an Extreme Learning Machine (ELM) model is combined with a variety of metaheuristic optimization techniques such as Particle Swarm Optimization (PSO), Differential Evolution (DE), Teaching Learning Based Optimization (TLBO), and Symbiotic Organism Search (SOS). These algorithms are helpful in creating ELM network weights and biases in the most effective way possible. The Teaching and Learning Based Optimization (TLBO) paradigm underpins most of the models and methodologies now in use for feature selection. To that end, this study sets out to identify the most effective machine learning model for making such predictions. The primary objective is to develop a smart technology that can predict the movement of stock market values based on financial time series and news headlines. Investors would be able to make more informed decisions based on the binary results of the proposed method. Processing time, error rate, and accuracy in forecasting the direction of stock price movement for five companies across sectors are used to evaluate the performance of the proposed LSTM-TLBO model.