Stock market prediction is a complex and dynamic challenge that has long intrigued researchers and traders. With advancements in Machine Learning (ML), data-driven methods have shown promise in forecasting stock price movements. This research introduces Candle Predict, an ML-based predictive model designed for the Indian stock market.

We explore several ML models, including Random Forest Regressor, XGBoost, and Long Short-Term Memory (LSTM) networks, to analyze historical stock data and forecast trends. Through extensive experimentation, LSTM networks proved most effective in capturing temporal dependencies and delivering accurate predictions.

The model is trained using five key stock parameters—Open, High, Low, Close, and Volume—sourced from Indian stock exchanges via the Yahoo Finance API. For evaluation, we employed an 80-20 train-test split and assessed model performance using Root Mean Square Error (RMSE). Results show that the LSTM model achieves an accuracy range of 87% to 94%, making it a dependable tool for short-term stock forecasting.

Our findings highlight the potential of deep learning techniques in financial prediction and emphasize the unique challenges of time-series data. Candle Predict serves as a practical and efficient solution for traders and investors aiming to make data-driven decisions in India’s volatile stock market.

This study contributes to the field of Financial Technology (FinTech) by demonstrating the effectiveness of ML in real-world market scenarios and offering a robust forecasting framework tailored to emerging market conditions.