Prediction of stock market using sentiment analysis and ensemble learning

People occasionally look to the stock market as an additional source of income. But investors face difficulties since stock market moves are inherently volatile and unpredictable. As a result, this study uses cutting-edge Deep Reinforcement Learning (DRL) approaches to increase the predictability of stock market patterns. A set of DRL models, namely Advantage Actor Critic (A2C), Proximal Policy Optimization (PPO2), and Soft Actor Critic (SAC), are utilized by utilizing data obtained from Yahoo Finance. The proposed Policy Adaptation with Trust Region Optimization (PACTRO), technique solve these issues by optimizing policy adaptation within limited trust regions. Key input components for prediction include technical indicators like the Price Chart, Moving Average Convergence Divergence (MACD), Bollinger Bands (BB), and Relative Strength Index (RSI). The created computational framework intends to deliver actionable insights, directing investors on optimal buy or sell decisions to maximize profit potential by synthesizing historical data within customized training and trading scenarios.•This study implements the Stock market prediction based on Deep Reinforcement Learning (DRL) using Technical Analysis and financials' statement.•DRL models like A2C, PPO2, and SAC, using data from Yahoo Finance, enhance investment decisions.•The proposed Policy Adaptation with Trust Region Optimization (PACTRO) method optimizes policy adaptation within trust regions.•Key technical indicators like ROC, MACD, Bollinger Bands, and RSI are crucial inputs for the prediction framework, helping guide optimal investment decisions.