Abstract: Fund investment style drift is a double-edged sword,in obtaining short-term excess returns,while also reflects the enormous drift risk.This paper chooses 8 open-end stock funds for the research sample founded in 2004,introduces skewed-t distribution to decipt new variables based on the peaks and heavy tails characters of fund investment style drift return series,adopt ARFIMA-HYGARCH-VaR model to measure fund investment style drift risk value,and with 5kinds of the Risk Metrics and GARCH family models to comparative analysis the VaR risk measures accuracy under skt distribution,and make fails backtest and dynamic quantile test for various VaR risk measure models,research results show that: In the different confidence level,the VaR value is more accurate based on ARFIMA-HYGARCH model;P erson goodness of fit test confirmes skt distribution can be well fitted for investment style drift return series.The conclusion of this study is undoubtedly of great theoretical and practical significance for controlling the investment style drift and standarding the fund products innovation and distribution.

Key words: fund investment style, investment style drift, style drift risk, SKT-HYGARCH-VaR model, model backtesting