标题: Quality Control Chart refresher [打印本页] 作者: DoubleDip 时间: 2011-7-11 19:32 标题: Quality Control Chart refresher
Specific LOS was to be able to interpret or read (cant remember) a quality control chart. Remember that the benchmark return will be the flat horizontal line with the two bands on top and on bottom. Just because a manager posts returns consistently above the benchmark flat line, it does not mean that you can reject the Null. The Null would be that the Portfolio Manager does not add value. The null is he sucks. If his returns consistently outperform the benchmark, but fall WITHIN the bands, you cannot reject the null. You cannot say he does not suck.
The same goes the other way and I could totally see CFAI asking a question where the returns are consistently below the benchmark, but within the bands. If they are below the benchmark, but within the bands, you cannot reject the null and say he does not suck. You fail to reject the null that he sucks because his low returns are still within the range bands.
You can reject the null that he does not add value if the returns are consistently above the top range band.作者: Darien 时间: 2011-7-11 19:32
I wouldn't say the null is that he sucks, but that he is just completely average. I think quality control charts can be used to reject the null because he is either good enough or bad enough to be statistically different from the horizontal line. I'd have to re-read it though as it's been a while作者: canadiananalyst 时间: 2011-7-11 19:32
good point...edited closer to your correct and accurate suggestion. thanks作者: Windjammer 时间: 2011-7-11 19:32
Sorry I have a question : what does it mean by Confidence Interval "WIDENS" in this context ?作者: lcw77 时间: 2011-7-11 19:32
alta168 Wrote:
-------------------------------------------------------
> Sorry I have a question : what does it mean by
> Confidence Interval "WIDENS" in this context ?
A wider confidence interval is decreasing the rejection range, like going from 90 to 95 percent confidence. This reduces your Type I error (false positives) but increases Type II error (failing to reject the null).
NO EXCUSES作者: mar350 时间: 2011-7-11 19:32
sbmarti2 Wrote:
-------------------------------------------------------
> I wouldn't say the null is that he sucks, but that
> he is just completely average. I think quality
> control charts can be used to reject the null
> because he is either good enough or bad enough to
> be statistically different from the horizontal
> line. I'd have to re-read it though as it's been
> a while
If the manager got bad result, and he will be out of the quality control chart too! (from the lower boundary.)
So we will reject the NULL, and does it mean we will fire the good manager??作者: skycfa 时间: 2011-7-11 19:32
I am confused.作者: bodhisattva 时间: 2011-7-11 19:32
H0 = Manager Adds No Value (he sucks)
HA = Manager Adds Positive Value (he's good)
Type I Error: Rejecting H0 when it is actually true.
Layman's terms: Investor may think that the manager adds value, when in fact, he sucks.
Type II Error: Failing to Reject H0 (AKA agreeing with H0) when it is actually false.
Layman's terms: Investor may think that the manager sucks, when in fact, he provided good value.作者: Roflnadal 时间: 2011-7-11 19:32
bpdulog Wrote:
-------------------------------------------------------
> alta168 Wrote:
> --------------------------------------------------
> -----
> > Sorry I have a question : what does it mean by
> > Confidence Interval "WIDENS" in this context ?
>
> A wider confidence interval is decreasing the
> rejection range, like going from 90 to 95 percent
> confidence. This reduces your Type I error (false
> positives) but increases Type II error (failing to
> reject the null).
A wider confidence interval is another way of saying a narrower significance level or in other words a narrower filter??? From 10 to 5% in a one tail test
