A Machine Learning Approach for Predicting Nifty 500 Index Constituents Using Historical Stock Data

This study is about using machine learning to figure out which companies will be part of the Nifty 500 index. The Nifty 500 index is like a list of companies on the National Stock Exchange and it helps us see how the market is doing in India. The way we do this is by looking at stock market data like the Open, High, Low and Close prices of stocks and how many stocks are being traded from 2015 to 2024. We also look, at which companies are already part of the Nifty 500 index. Then we use this data to come up with some indicators like how well a stock does over time how volatile it's how easy it is to buy and sell and some other metrics that help us rank companies. The main goal of the Nifty 500 index study is to rank companies based on how they are to be included in the Nifty 500 index. We use the Nifty 500 index to see which companies are most likely to be part of it. The XGBoost model is really good. It works well with structured data. We have to make sure we do the evaluation correctly. We use walk-forward validation. The XGBoost model is correct 88% of the time when we look at past results. For example, when we used it to make predictions for April 2026 it was correct 74.4% of the time. The results show us that the amount of money available to buy and sell a stock and how well the stock does over time are things to consider when deciding if a stock should be part of the index. The idea we are proposing can also be used for things like predicting what will happen with money and picking stocks for a portfolio. The Nifty 500 index and the stocks that are part of it are very important for people who invest money and for market analysts. The framework we made using machine learning can help investors make decisions. We also learned more about what makes a stock part of the index. This is useful for the XGBoost model and, for people who use it.