MUMBAI, India, July 7 -- Intellectual Property India has published a patent application (202641077924 A) filed by Dr. Chaitanya S Kittur; Dr. Manali Agrawal; and Shreedhar Deshmukh on June 24, 2026, for A Machine Learning Framework For Real-Time Portfolio Rebalancing And Risk Control.
Inventors include Dr. Chaitanya S Kittur; Dr. Manali Agrawal; and Shreedhar Deshmukh.
The application for the patent was published on July 03, 2026, under issue no. 27/2026.
Abstract: A Machine Learning Framework for Real-Time Portfolio Rebalancing and Risk Control The present invention discloses a Machine Learning Framework for Real-Time Portfolio Rebalancing and Risk Control designed to optimize investment performance, minimize financial risks, and dynamically adapt to rapidly changing market conditions. Traditional portfolio management approaches rely on periodic rebalancing strategies and static risk assessment models, which often fail to respond effectively to market volatility, sudden price fluctuations, and emerging economic events. The proposed framework integrates advanced machine learning algorithms, predictive analytics, and real-time data processing techniques to continuously monitor market conditions, evaluate portfolio performance, and automatically execute intelligent rebalancing decisions. The system collects and processes multi-source financial data, including stock prices, trading volumes, market indices, macroeconomic indicators, news sentiment, social media trends, and investor behavior patterns. A data preprocessing module performs normalization, feature extraction, noise reduction, and anomaly detection to generate high-quality input for predictive modeling. The machine learning engine employs ensemble learning models, deep neural networks, reinforcement learning agents, and time-series forecasting techniques to predict asset returns, volatility levels, liquidity risks, and market trends. Based on these predictions, a portfolio optimization module determines the optimal asset allocation by balancing expected returns with predefined risk constraints. The framework further incorporates a real-time risk control engine that continuously calculates risk metrics such as Value-at-Risk (VaR), Conditional Value-at-Risk (CVaR), drawdown probability, Sharpe ratio, beta coefficient, and market exposure. When predefined risk thresholds are exceeded, the system automatically generates corrective actions, including portfolio rebalancing, asset diversification, position resizing, hedging recommendations, and capital preservation strategies. Additionally, a reinforcement learning-based decision layer continuously learns from historical outcomes and live trading environments to improve future allocation decisions and risk management effectiveness
Disclaimer: Curated by HT Syndication.
