Abstract: There is a bias in which market timing ability of funds is negative among all models estimating the performance of funds.This paper constructs the skewness adjustment conditional quadratic model based on the quadratic regression model of Treynor and Mazuy.By the empirical analysis of Chinese security investment funds,we prove that this model is more effective and erases the bias of negative market timing ability.Because there are series relations among the excess returns of security-investment funds,they are inefficient estimators using OLS or GLS for the time series of the excess returns of security investment funds.The paper evaluates the performance of Chinese funds using seemingly unrelated regression equations (SURE),and analyzes comparatively the estimators between OLS (GLS) and SURE.The result is the conclusion of SURE is much more efficient than that of OLS (GLS).

Key words: security-investment funds, market timing ability, skewness, SURE