THARUN RATHOD

CH 02 AI & ML

NIFTY-50 Investment Intelligence

21 years of NIFTY-50 data with Hidden Markov Models detecting bull/bear regime transitions

  • XGBoost price prediction with SHAP explainability showing which features drove each prediction
  • Portfolio optimization via Modern Portfolio Theory, Value-at-Risk analysis
  • HMM detects regime shifts 3-5 days before they appear visually on price charts

Details

  • HMM trained on return distributions detects regime shifts 3-5 days before they are visually apparent on price charts
  • SHAP values surface which features (RSI, MACD, volume delta) drove each prediction - explainability built in, not bolted on
  • Finance
  • Python
  • XGBoost
  • SHAP
  • Streamlit