B*******t
发帖数: 135 | 1 有两个portfolios：P1和P2。他们的1day的5%的VaR分别是2m和3m，
那么把这两个portfolio合起来作为一个大的porfolio来考虑，大portfolio的1day 5%
的VaR是多少？
好像这个问题还挺复杂的。两个portfolio的价值，V1和V2。如果仅仅知道 DeltaV1和
DeltaV2的correlation，可以得到最终答案么？还是说必须DeltaV1和DeltaV2的joint-
distribution？
先提前谢谢大牛们指教了！ |
B*******t
发帖数: 135 | |
a***m
发帖数: 74 | 3
joint-
For the combined portfolio, you can first calculate the volatility. This
require info including positions, correlations of the returns, and
volatility, which you have. Then you can estimate VAR use volatility and
position.It's straight-forward.
【在 B*******t 的大作中提到】
: 有两个portfolios：P1和P2。他们的1day的5%的VaR分别是2m和3m，
: 那么把这两个portfolio合起来作为一个大的porfolio来考虑，大portfolio的1day 5%
: 的VaR是多少？
: 好像这个问题还挺复杂的。两个portfolio的价值，V1和V2。如果仅仅知道 DeltaV1和
: DeltaV2的correlation，可以得到最终答案么？还是说必须DeltaV1和DeltaV2的joint-
: distribution？
: 先提前谢谢大牛们指教了！
|
p******i
发帖数: 1358 | 4 var12=sqrt(var1^2+var2^2+2rhoVar1Var2)
is here 2rho or rho I forgot
someone correct me pls if I am wrong |
B*******t
发帖数: 135 | 5
The var you mentioned here is VARIANCE......
I was asking about Value-at-Risk.
【在 p******i 的大作中提到】
: var12=sqrt(var1^2+var2^2+2rhoVar1Var2)
: is here 2rho or rho I forgot
: someone correct me pls if I am wrong
|
B*******t
发帖数: 135 | 6 I considered the problem in a mathematical way and thought the joint-
distribution of the DeltaV2 and DeltaV2 should be given to figure out the
VaR of the big portfolio.
Let's start from the mathematical (probability) meaning of VaR. Assume the
portfolio value today is V and the increase by tomorrow is DeltaV, then the
5% VaR is the minus value of the 5% percentile of DeltaV. In other words, if
the %5 percentile of DeltaV is -2, then 5% Value-at-Risk = 2.
Therefore, the original VaR problem can |
p******i
发帖数: 1358 | 7 yeah,right
I need you to teach me what is VAR
【在 B*******t 的大作中提到】
: I considered the problem in a mathematical way and thought the joint-
: distribution of the DeltaV2 and DeltaV2 should be given to figure out the
: VaR of the big portfolio.
: Let's start from the mathematical (probability) meaning of VaR. Assume the
: portfolio value today is V and the increase by tomorrow is DeltaV, then the
: 5% VaR is the minus value of the 5% percentile of DeltaV. In other words, if
: the %5 percentile of DeltaV is -2, then 5% Value-at-Risk = 2.
: Therefore, the original VaR problem can
|
c**a
发帖数: 316 | 8 VaR = Value at Risk
如果 高斯 分布 算 方差就够了。
献丑了。
【在 p******i 的大作中提到】
: yeah,right
: I need you to teach me what is VAR
|
B*******t
发帖数: 135 | 9 Don't be offended, man. I just thought I had not made it clear enough.
I read your previous post again. I am sorry was wrong: you were talking about Value-at-Risk. I thought you were talking about VARIANCE because you put it in a looks-too-simple way. You conclusion was based on a series of strong assumptions, which you probably thought are too obvious and straightforward to point out.
【在 p******i 的大作中提到】
: yeah,right
: I need you to teach me what is VAR
|
B*******t
发帖数: 135 | 10 I reread the VaR chapter in John Hull's book.
Seems like he made a further assumption that mu1=mu2=0. Then the problem is
much easier since now you have
Value-at-Risk = c * STD,
where c is a constant determined by your confidence level.
【在 B*******t 的大作中提到】
: Don't be offended, man. I just thought I had not made it clear enough.
: I read your previous post again. I am sorry was wrong: you were talking about Value-at-Risk. I thought you were talking about VARIANCE because you put it in a looks-too-simple way. You conclusion was based on a series of strong assumptions, which you probably thought are too obvious and straightforward to point out.
|
p******i
发帖数: 1358 | 11 the assumption is not strong at all
the only assumption is that you use IGARGH(1,1) with normal residual which
is sort of industry standard known as RISKMETRICS model
also VAR is a wrong and useless concept anyway, so why bother to find the '
theoretically sound' way
about Value-at-Risk. I thought you were talking about VARIANCE because you
put it in a looks-too-simple way. You conclusion was based on a series of
strong assumptions, which you probably thought are too obvious and
straightforwar
【在 B*******t 的大作中提到】
: Don't be offended, man. I just thought I had not made it clear enough.
: I read your previous post again. I am sorry was wrong: you were talking about Value-at-Risk. I thought you were talking about VARIANCE because you put it in a looks-too-simple way. You conclusion was based on a series of strong assumptions, which you probably thought are too obvious and straightforward to point out.
|
j*****o
发帖数: 6 | 12 If you said VaR is useless clearly you've never really worked in the market
risk business. In most cases, it is still the single most important measure
people looking at when speaking about market risk. A strong evidence is how
regulators still place great emphasis in VaR.
That said, one is easy to see that VaR is far from perfect theoretically.
However, no models even the Black-Scholes are perfect in the finance world.
People use them because emperically they give satisfactory results.
In risk
【在 p******i 的大作中提到】
: the assumption is not strong at all
: the only assumption is that you use IGARGH(1,1) with normal residual which
: is sort of industry standard known as RISKMETRICS model
: also VAR is a wrong and useless concept anyway, so why bother to find the '
: theoretically sound' way
:
: about Value-at-Risk. I thought you were talking about VARIANCE because you
: put it in a looks-too-simple way. You conclusion was based on a series of
: strong assumptions, which you probably thought are too obvious and
: straightforwar
|
