开发者:上海品职教育科技有限公司 隐私政策详情

应用版本:4.2.11(IOS)|3.2.5(安卓)APP下载

shirleygong2016 · 2022年03月13日

能不能解释下b

NO.PZ2018122701000030

问题如下:

Let X be a random variable representing the daily loss of your portfolio. The "peaks over threshold" (POT) approach considers a threshold value, u, of X and the distribution of excess losses over this threshold. Which of the following statements about this application of extreme value theory is correct?

选项:

A.

To apply the POT approach, the distribution of X must be elliptical and known.

B.

If X is normally distributed, the distribution of excess losses requires the estimation of only one parameter, β, which is a positive scale parameter.

C.

To apply the POT approach, one must choose a threshold, u, which is high enough that the number of observations in excess of u is zero.

D.

As the threshold, u, increases, the distribution of excess losses over u converges to a generalized Pareto distribution.

解释:

D is correct.

考点: Extreme Value

解析: The distribution of excess losses over u converges to a generalized Pareto distribution as the threshold value u increases.

The distribution of X itself can be any of the commonly used distribution: normal, lognormal, t, etc., and will usually be known. The distribution of excess losses requires the estimation of two parameters, a positive scale parameter β and a shape or tail index parameter ξ. One must choose a threshold u that is high enough so that the theory applies but also low enough so that there are observations in excess of u.

不太理解B的意思。 应该有两个参数 一个是tail factor 一个是positive sclae


2 个答案

DD仔_品职助教 · 2022年11月21日

嗨,努力学习的PZer你好:


只是知道是正态分布,正态分布的尾部参数又不是一定的呀,所以还是需要知道俩个factors

----------------------------------------------
就算太阳没有迎着我们而来,我们正在朝着它而去,加油!

DD仔_品职助教 · 2022年03月13日

嗨,爱思考的PZer你好:


同学你好,

你已经解释出来了为什么B是不对的呀~

B说如果X是趋于正态分布的话,在估计他的极值时只需要一个factor。

这个是不对的,需要两个factor,如下图:


----------------------------------------------
努力的时光都是限量版,加油!

映昔_lei · 2022年11月21日

请教下,如果B选项说的他已经趋于正态分布了,那么尾部参数不也是已知与正态一直了吗?所以只差一个scale?

  • 2

    回答
  • 0

    关注
  • 489

    浏览
相关问题

NO.PZ2018122701000030 问题如下 Let X a ranm variable representing the ily loss of your portfolio. The \"peaks over threshol" (POT) approaconsirs a thresholvalue, u, of X anthe stribution of excess losses over this threshol Whiof the following statements about this application of extreme value theory is correct? To apply the POT approach, the stribution of X must ellipticanknown. If X is normally stribute the stribution of excess losses requires the estimation of only one parameter, β, whiis a positive scale parameter. To apply the POT approach, one must choose a threshol u, whiis high enough ththe number of observations in excess of u is zero. the threshol u, increases, the stribution of excess losses over u converges to a generalizePareto stribution. is correct. 考点: Extreme Value 解析: The stribution of excess losses over u converges to a generalizePareto stribution the thresholvalue u increases. The stribution of X itself cany of the commonly usestribution: normal, lognormal, t, etc., anwill usually be known. The stribution of excess losses requires the estimation of two parameters, a positive scale parameter β ana shape or tail inx parameter ξ. One must choose a thresholu this high enough so ththe theory applies but also low enough so ththere are observations in excess of u. 老师,这个应该是什么要求呢,忘了==

2023-01-18 23:42 1 · 回答

NO.PZ2018122701000030

2022-01-22 18:15 1 · 回答

NO.PZ2018122701000030

2021-08-07 21:51 1 · 回答

If X is normally stribute the stribution of excess losses requires the estimation of only one parameter, β, whiis a positive scale parameter. To apply the POT approach, one must choose a threshol u, whiis high enough ththe number of observations in excess of u is zero. the threshol u, increases, the stribution of excess losses over u converges to a generalizePareto stribution. is correct. 考点: Extreme Value 解析: The stribution of excess losses over u converges to a generalizePareto stribution the thresholvalue u increases. The stribution of X itself cany of the commonly usestribution: normal, lognormal, t, etc., anwill usually known. The stribution of excess losses requires the estimation of two parameters, a positive scale parameter β ana shape or tail inx parameter ξ. One must choose a thresholu this high enough so ththe theory applies but also low enough so ththere are observations in excess of u. n增大,POT不是更趋近于广义极值理论吗?POT本身就是广义帕累托,怎么能说趋近呢?

2021-01-18 20:43 2 · 回答