什么时候是双尾什么时候是单尾

问题如下：

A risk manager is backtesting a company’s 1-day 99.5% VaR model over a 10-year horizon at the 95% confidence level. Assuming 250 trading days in a year and the daily returns are independently and identically distributed, which of the following is closest to the maximum number of daily losses exceeding the 1-day 99.5% VaR in 10 years that is acceptable to conclude that the model is calibrated correctly?

选项：

19

25

35

39

解释：

中文解析：

考察backtesting VAR的公式：

p是尾部概率=0.5%

T是总共观察值=250*10=2500

z=1.96

代入数据即可得出x=19.4

取整=19

A is correct. The risk manager will reject the hypothesis that the model is correctly calibrated if the number x of losses exceeding the VaR is such that:

where p represents the left tail level and is equal to 1-0.995, or 0.5%; and T is the number of observations = 250*10=2500. And z = 1.96 is the two-tail confidence level quantile, given a confidence level of 95%.

If

then, x = 19.4.

So, the maximum number of exceedances would be 19 to conclude that the model is calibrated correctly.