- 金币:
-
- 奖励:
-
- 热心:
-
- 注册时间:
- 2001-4-26
|
|
楼主 |
发表于 2007-5-15 19:26
|
回复 #38 野狐禅 的帖子
yes,I am drawing I(t).usually we use I(t)^2,that means Var(R(t)|F(T)).
here,
Ut=E(Rt|Ft), (Ft means all informations at "t" time,"|"means under this condition)
It^2=Var(Rt|Ft)=E[(Rt-Ut)^2|Ft]=volatility=conditional variance
all things are time-change,so we get time series.
we have hundreds of models to draw It,the frist one is "ARCH" by Engle, who got Nobel Prize in 2000 for it.the Pop model at now is Garch,but it is not good enough,peoples are keeping on fixing it,the latest is MCMC-Garch."极差修正GARCH” is a simple way to get MCMC-Garch.
the abnormal volatility is this: 1.we get "It" or "normal volatility" by MCMC-garch(or any other model you like),2.define "abnormal It",it's not "Rt-Ut",we need some way to distinguish it
from the total volatility-series,such as "MSN algorithm".
[ 本帖最后由 无聊嘛 于 2007-5-15 19:38 编辑 ] |
|
|