8 science and technology themed funds in 2020 was selected as samples. Due to the peak and thick tail characteristics of fund returns, POT model of Extreme Value Theory was used to calculate VaR and CVaR risks. Comparing the effects of RAROC based on VaR and CVaR on fund performance evaluation,the results found that VaR and CVaR calculated based on POT model of the Extreme Value Theory can well reflect the tail risk of fund returns. The optimal threshold of the POT model selected by combining the empirical method and the graphical method can obtain effective VaR and CVaR. The RAROC indicator based on CVaR is more stable and effective in fund performance evaluation than the RAROC indicator based on VaR, and it can positively reflect fund returns and avoid risks.