|
- UID
- 223352
- 帖子
- 216
- 主题
- 39
- 注册时间
- 2011-7-11
- 最后登录
- 2013-8-23
|
2#
发表于 2011-10-4 03:41
| 只看该作者
1) You can find betas without needing any kind of theory, such as CAPM.
2) Three months of monthly data is not sufficient to properly calculate a beta. This is because it is built upon linear regression techniques, which typically require a larger sample size.
3) There are multiple methods to calculate the beta. Best is to regress the excess returns of the security against the market. Under some conditions, the formula for this is correlation(security-rf,mkt-rf)*stdev(security-rf)/stdev(mkt-rf). If you can't figure out how to calculate excess returns, standard deviations, or correlations, then you should look them up on wikipedia (beta is also on wikipedia).
4) If you want to check answers, I got stdev(A-rf)=8.8%, stdev(B-rf)=4.6%, stdev(mkt-rf)=3.1%, cor(A-rf,mkt-rf)=-0.23,cor(B-rf,mkt-rf)=-1. Beta(A)=-0.64 and Beta(B)=-1.49.
5) Technically, you would regress the excess log returns against excess log returns, project out to your horizon, convert to arithmetic returns, and then recalculate for the arithmetic beta, but since you are only working with arithmetic returns over one horizon you are okay.
Edited 1 time(s). Last edit at Wednesday, May 4, 2011 at 03:32PM by jmh530. |
|
