IV - Probability Distribution-4 Flashcards
区分:
discrete random variable
continuous random variable
前者:结果数量是可数的,即使有无限多个。对每一个结果有一个可以衡量的概率
一个月中下雨天数
后者:结果是不可数的,即使有上下界限。对任意一个结果,其概率为零,只能求得区间概率。
一个月降雨量(1-1000毫升)
似乎主要是一个确定单位的问题
3 ways of discribing
probability distribution
表达方式,描述的是什么
- probability function: for discrete, p(x)
- probability density function/pdf: for continuous, f(x)
- cumulative distribution function/cdf F(x) = P (X <=x)
binomial distribution:
Bernoulli trial
equation
区分Bernulli实验
和二项式分布
相同点1,不同点2
- 相同:都是只有两种outcome
- 不同:1.Bernulli只做一次实验,二项式做n次
- 两者期望值、方差不同
continuous uniform distribution
有一个怎样的特点
公式表示
一条水平线段
multivariate distribution
需要知道几个元素?3
- n个mean
- n个variance
- n(n-1)/2个correlation
normal distribution
性质4条
confidence interval
需要记住的4组值
辨析
正态分布中,+-1.96标准差,得95%置信区间
根据Chebyshev不等式,+-1.96标准差,得75%置信区间
Chebyshev是普世适用,得最低值(至少75%)
标准分布图形收得更紧
standard (unit) normal distribution
definition, and
Z value
标准正态分布,以Z表示
μ=0, s=1
Z value = (X-μ) / s
standarizing a random variable:
equation
shortfall risk
definition
the risk a portfolio value will fall below the
客户要求的最低回报
Roy’s
Safety First Ratio
equation
及推论
SFRatio 越大者越符合要求
比较
SFRatio
&
Sharpe Ratio
当客户要求最低回报率就是无风险回报率时,
SF ratio = Sharpe ratio
lognormal distribution
1。定义
2。图形左右偏
3。极限值
4。应用
- if LnX is normal, then X is lognormal
- right / positively skewed
- 0 - 正无穷
- describe asset price
normal distriution
lognormal distribution
分别用来描述
- returns
- share and asset price
if continuously compounded returns are normally distributed
then asset prices are lognormally distributed
lognormal distridution:
已知连续复利回报HPR,求名义利率r
er - 1 = HPR
er = HPR + 1
r = ln (HPR+1)
Monte Carlo Simulation
描述1 点 区别 2点
Historical Simulation
- M对风险因素进行虚拟建模,H是对过去实际发生的风险因素进行统计
- M的缺点:一是复杂,二是需根据输入的分布参数来进行模拟
- H的缺点:未来的风险因素不一定等于过去的,因此H不能解决“what if”的问题。选取的时间段若未包含罕见的极端情况,则无法反映在统计结果中
For a binomial random variable with five trials,
and a probability of success on each trial of 0.5,
the distribution will be:
Symmetric.
Other than P=0.5, the distribution would be asymmetric
The total number of parameters that fully characterizes a multivariate normal distribution for the returns on two stocks is :
5.
two means, two variables, and one correlation
Monte Carlo simulation
实质是
providing a distribution of possible solutions to complex functions