Stock Buy and Sell Predictions Based on Technical Analysis and News Sentiment Using Random Forest

The fluctuations in stock market prices are a compelling subject for examination due to their significant profit potential. Market participants utilize stock transaction data and its derivatives, such as technical analysis indicators, to forecast stock prices. Sometimes technical analysis is combined with stock news sentiment to strengthen predictions. An automated system is required for more objective and expedited results due to the subjectivity of technical analysis predictions and news sentiment. Therefore, this study aims to develop a stock buy and sell prediction model by combining historical data and news sentiment. The technical analysis indicators used are: Bollinger Bands (BB), Exponential Moving Average (EMA), Simple Moving Average (SMA), Relative Strength Index (RSI), and Moving Average Convergence/Divergence (MACD). News data is categorized into positive, negative, or neutral sentiment. The fusion of historical data and news sentiment is used for buy and sell prediction input with random forest. The results of buy and sell predictions with random forest are compared with other methods, namely SVM and LSTM. The results show that the proposed method, random forest, gets an accuracy of 80.74%, better than SVM and LSTM.