Abstract: Performance measurement is one of the most important issues in the research of mutual funds. The problems of performance evaluation and tests in the open-end mutual funds are studied in this paper, using daily returns. Conditional AutoRegressive Expectile (CARE) models are creatively introduced into the problem of evaluation of mutual funds' performance. Firstly, asymmetric least squares (ALS) method is applied to estimate the parameters in those CARE Models, and then the results are used to create autoregressive VaR model and conditional ES model to calculate the values of VaR and ES of our sample funds. Secondly, the values of VaR and ES are used to conduct risk-adjustment on the standard deviation, and thus the amended Sharpe ratios are obtained, which are based on VaR and ES. Finally, in empirical study, 56 domestic open-end funds in China are selected as samples, from 2005 to 2011. Empirical analysis are made on the evaluation and ranking of three measures of performance, including the traditional Sharpe ratio, VaR-based Sharpe Ratio and ES-based Sharpe ratio. The results prove CARE models can measure extreme risk much more accurately and thus can be very feasible to the evaluation and test in mutual funds.

Key words: Sharpe ratio, asymmetric least squares, conditional autoregressive expectile models, VaR, ES