Advancing Stock Price Prediction with Hybrid Machine Learning: A mixer of Sentiment and Financial Indicators

Accurately predicting stock prices is a complex task in the financial markets, as the stock market is the place where a small improvement in prediction accuracy can lead to significant financial gains. The price of the stock market is highly volatile due to many economic, social, and political reasons, which is challenging for investors to analyze all together. To cope with this problem, we proposed a hybrid machine learning approach that combines financial indicators with news sentiment analysis. The given approach uses four methods: penalized regression, support vector machine (SVM), natural language processing (NLP), and long short term memory (LSTM) networks. In a penalized regression, the system performs feature selection. The system will perform news filtering and sentiment analysis through SVM and NLP techniques, and in the Long Short-Term Memory (LSTM) system, it predicts stock prices. The model uses technical indicators like OHLCV, ROE, ROIC, ROA, D/E, moving averages, RSI, and Bollinger Bands along with sentiment scores from relevant financial news articles. As a result, this hybrid model significantly increases forecast accuracy compared to traditional models such as ARIMA and Random Forest.