The Improvement Of VaR Calculation And Used In Evaluation Of Mutual Fund

Funds as an invest instrument has made significant advance through just a fewyears. Scientific fund appraisal does well to its healthy development. It is importantand meaningful to find a reasonable fund appraisal. In the capital market, high rateof interests always accompanied with high risk, so it is more credible to think moreabout risk when we appraise funds. We usually use standard deviation ,βcoe?cient,VaR value to measure risk. The classic Treynor measurement, Sharp measurement andJensen measurement depends on the three risk value. In this paper, I used VaR to revisethe traditional sharpe ration and make an analysis of the open-fund's performance. Ialso compared the traditional sharpe ration with the revised sharpe ration and got aconclusion that the revised sharpe ration is more appropriate to measure the risk offunds. There are two ways in computing of VaR value, which is parametric methodinclude and nonparametric method. In the computation of VaR value, I chose modelbased on extreme instead of normal model, which is the classic parametric model. Ialso add exponentially weight in the model of historical data. Finally, I use the back-testing of Kupiec failure to test validity of VaR.In the second chapter of this paper, I chose 50 mutual funds on the internet andcompute the return of them. I find that all of the mutual funds tend to approach thenormal distribution. But all of their kurtosis index and skewness index of return di?erwith the normal distribution.The third chapter is about the computation of VaR value, which is the most im-portant of this paper. Based on the distribution of return, I can not use the classicparametric method-the model of variance and coe?cient variance. So I chose POTmodel based on the extreme theory. As the classic model of historical has some de-faults, I add exponentially weight to revise it. I use the back-testing of Kupiec failureto test validity of VaR and I get the conclusion that VaR model based on the extremetheory is valid in high confidence levels and invalid in low confidence levels, the model of revised historical is on the contrary.In the forth chapter I compute and comprise di?erent fund appraisal model. Thefifth chapter mentions the conclusion, innovation and defaults of this paper.