Research On UHF Volatility Models

Financial asset volatility is an important factor for asset evaluation, hedging stategies and risk management.Withing the devolopment of computer technology, thousands of data can be stored and computed. Data which record every transaction called UHF.UHF take advantage on studying financial microstructure and risk management.But the time is stochastic when tractions happen and the data is irregularly spaced between tractions.In the past, ARCH model and SV model were widespread in evaluating volatility, which they are applying in regular interval, but this way may ingnore asset price influenced by information. So, I introduced a new model called ACD and it's augment family, and studied the UHF-GARCH which proposed by Engle, finally I reaserched the volatility for irregular spaced data.In the demonstration part, I analyse the statistic characteristic, stationarity and dependence of the UHF and daily data for ZHONGXIN BANK.Then I modeled the daily data and UHF data by GARCH(1,1) and UHF(1,1) respectively.The empirical result indicate UHF-GARCH has smaller MAE and RMSE than GARCH.So UHF-GARCH model perform better than GARCH on forcasting volatility.From the result of the second equation of UHF-GARCH, we can find that the factor of 1/xi is negative., but it's Z statistic not obvious.In the volatility equation.That's indicate long trade duration don't mean bad news for ZHONGXIN.In additon, the factor of trade duration xi is positive, that's long trade duration means there is no news and it's volatility become smaller.