We present some shortcomings about the traditional models evaluating the timing ability of mutual fund managers and propose a new model to evaluate the mutual fund managers’timing ability to tendancy change. Using this model and net asset value data , we can perceive whether the mutual fund managers can predict the future mov2 ing direction of the market when it experiences a significant tendancy change. Monte Carlo simulations indicate that our model does fulfill its task while other traditional models such as Treynor-Mazuy model and Henrikson2Merton model fail . Applying our approach to empirical study of the Chinese mutual funds , we get some interesting facts about the Chinese mutual fund managers