A Survey on Regime-Aware Portfolio Analytics and ML-Based Strategy Evaluation for Indian Equity Markets

Authors:

Mrs. Beena K¹, Jayaditya Dev², Durgashree M³, Aryaman Tiwari³, Chirag T⁵

Assistant Professor, Dept of CSE, KSIT, Karnataka, India¹

Student, Dept of CSE, KSIT, Karnataka, India^2-5

Abstract – The increasing complexity and volatility of financial markets have created a growing demand for intelligent systems capable of accurate stock prediction, adaptive trading, and efficient risk management. Traditional statistical methods and many existing financial platforms primarily focus on visualization and isolated backtesting, often lacking integrated regime-aware and risk-aware analytical capabilities. This survey reviews recent advancements in machine learning, deep learning, and reinforcement learning techniques applied to stock market prediction, algorithmic trading, market regime detection, portfolio optimization, and financial stress testing. The study examines widely used models such as Long Short-Term Memory (LSTM), Gated Recurrent Unit (GRU), Deep Neural Networks (DNN), Convolutional Neural Networks (CNN), Autoencoders, and Deep Reinforcement Learning (DRL). Additionally, emerging approaches involving sentiment analysis, cross-market data integration, and regime clustering are discussed. Key challenges including backtest overfitting, data non-stationarity, and model generalization are also analyzed. The reviewed literature indicates that hybrid machine learning approaches can significantly improve forecasting accuracy and trading performance compared to traditional techniques, while challenges related to robustness and real-time adaptability remain important research areas.

KEY WORDS: Stock Market Prediction, Algorithmic Trading, Machine Learning, Deep Learning, Reinforcement Learning, Market Regime Detection, Financial Risk Management, Sentiment Analysis, Portfolio Optimization