ANN-GJR-GARCH Model Applied in the Evaluation of High Frequency Taiwan Index Future Intra-day Data

• Main Authors: Ying-Yi Li, 李盈儀
• Other Authors: Wo-Chiang Lee
• Format: Others
• Language: zh-TW
• Published: 2004
• Online Access: http://ndltd.ncl.edu.tw/handle/42378670793040735641

碩士 === 真理大學 === 財經研究所 === 92 === The paper used per 10 secs Taiwan stock index futures intra-day high frequency data and tried to apply three variation GARCH-type volatility models,named GARCH,GJR-GARCH,ANN-GJR-GARCH VaR models. In The last,we used mean、percentage volatility、mean distance of VaR，RMS，error validation and back-test to evaluate VaR models. Results show that ,GARCH model,GJR-GARCH and ANN-GJR-GARCH model have the same intra-day volatility clustering. However, comparison of intra-day and inter day data,we find different volatility clustering figures. The volatility of inter day is higher than intraday data . It also shows that to make decision by inter aday data bring a high risk. To test the stability of GARCH-type VaR by Wilcoxon sign test，we find the ANN-GJR-GARCH VaR model is most unstable between all VaR models. But GARCH-VaR model has the best stability.