Comparative Analysis of Stock Price Forecasting using MTF Image Transformation: Evaluation of Random Forest and CNN Models

Accurate stock price forecasting is a pivotal endeavor in financial markets, essential for informed investment decisions and effective economic strategies. Traditional models such as CAPM, Black-Scholes, ARIMA, and GARCH often struggle to capture the complex, non-linear dynamics inherent in market data. This study presents a comparative analysis of deep learning-based approaches to predict trends in the financial time series. We propose a novel methodology that transforms raw financial data into image representations using the Markov Transformation Field (MTF) alongside a specialized image creation algorithm tailored for financial analysis. These images are then classified into the Buy, Sell, and Hold categories. The approach is evaluated on a dataset comprising twenty stocks from the Indian financial sector, offering a comprehensive assessment of model performance on a substantial scale. To be precise, our experiments suggest an F1 score of 73%. These numbers demonstrate that the proposed deep learning-based technique achieves promising accuracy, highlighting its potential advantages over traditional forecasting methods in capturing the behavior of financial markets.