Stock market risk prediction of Indian stocks using Kalman filter: Towards automation of finance industry

The Stock market is known for its unpredictability and time-varying nature in real life scenario, thus stock market can be treated as stochastic dynamical system. In this paper, the stock prices of two Indian companies (State Bank of India and Mahindra & Mahindra) and their volatility in the Indian market is investigated by estimating the time-varying coefficient ‘Beta’ $(\beta_{t})$ of Capital Asset Pricing Model (CAPM). The ‘Beta’ estimate is obtained in this work by implementing discrete Kalman filter in recursive estimation mode using MATLAB software platform. The estimated ‘Beta’ is further used to compute the return on the stock price and thereby compare the same with actual market return data for different values of noise variance. The investigation carried out in this work is a step forward towards an automation of finance industry that warrants the judicious use of mathematical models and associated prediction algorithms to aid investors in taking proper decisions.