Functional Data Analysis For Volatility

CSI300Stock Index Future is the first financial future goods. The appearance of CSI300Stock Index Future is very important to perfect the structure of our financial market and give our investor a tool of risk management. Because of the margin trading and marking-to-market in the future market, the volatility of Future contract will multiple amplify the gain or loss of the investor. Hence it is necessary to study the volatility of CSI300Stock Index Future.In my study, we adopt a method named Functional Data Analysis, aiming at modeling data and extract information about volatility for situation where repeated realizations of the volatility process are available. The sample in my empirical analysis part is high frequency data of139trade days from Oct.18th2010to May202011. Using Functional Data Analysis, we found a "Noon Effect" of the mean processes of the volatility and four common factors. Then we fit the four loadings using vector auto-regressive (VAR) model and predict the volatility processes90days after. Our results show that the model using KL decomposition and AR(3) model might yield the best prediction performance.Up to our knowledge, this might be the first paper that studies the volatility of CSI300Stock Index Future using Functional Data Analysis for modelling and prediction. Our finding might benefit both the investors and the government.