Volatility - comparision of predictive models in Indian market

Abstract

Abstract: 1) Introduction and context description: Studies have been done which capture the volatility and the behaviors of the predictive models in the Indian Equity market. But these studies use data from the time when option trading was not allowed in India. Implied Volatility as a predictive model of market volatility is therefore still untested in India. Research in this area suggests a better prediction of market short term volatilities from the option model viewpoint. After almost three years after the introduction of option trading in the Indian equity market, we can assume a greater deal of maturity in the market. If the research does not conform to the well know findings, we can expect two divergent views from it. Either the options market is still not mature or the models that predict western markets are not good enough for the Indian market and there is a need for specific predictive models. 2) Research questions: 1. Determine the GARCH Model to be used for the specific Indian Market. 2. Evaluating the correct indicator of actual volatility for the Indian Market. The ideal measure is realized volatility, but with the possibility of data not forthcoming, the alternate measure could be used. This alternate measure can be on of the "Extreme Value volatility estimators" based on High, Low, Open and Close prices. 3. To ascertain which model gives the better prediction for volatility under different conditions. 3) Methodology: 1. Data and study done from previous similar research papers were used to find the effective way of calculating realized volatility. 2. Collection of Data: Option price data and the index closing price data was picked up from NSE website. VOLATILITY TRADING • COMPARISON OF PREDICTIVE MODELS IN INDIAN MARKET 3. Implied Volatility: NIFTY options were used to calculate the implied volatility. The NIFTY index is much more liquid than the individual stocks in the option market and hence a better predictor of volatility. 4. Range of Data: Sep 2001 - Oct 2005. 5. Application of the model: Option model is used to get the measure of implied volatility of the NSE index. GARCH provides another measure of future volatility. 6. Evaluation of the Results: Comparative analysis was done to evaluate the performance of the two predictive models. An important criterion was to ascertain the difference in level of predictability of the two models. 4) Findings • Implied volatility is a biased and efficient estimator of realized volatility • In comparison, GARCH is unbiased but does not follow the trends in realized volatility • GARCH and Implied volatility model have no significant differences in error 5) Limitations of the study In the study we have used monthly comparisons. Volatility estimates vary everyday and a monthly estimate fails to capture these movements. 6) Scope for further work • Use high frequency data to get a better measure of realized volatility. • Compare the time series change in volatility for the options expiring on the same day. 7) Key words GARCH, Implied Volatility, realized volatility Learning from the IP process: We developed an understanding of the following while working on this IP: • Analysis of large volume of data • Challenges in Data Mining Use of descriptive statistics for validation