[blml] 25 something ?

[John Probst]

----- Original Message ----- 
From: "Steve Willner" <willner at cfa.harvard.edu>
To: <blml at rtflb.org>
Sent: Wednesday, December 06, 2006 3:29 AM
Subject: Re: [blml] 25 something ?


>> From: "John Probst" <john at asimere.com>
>> As i said, read law 15A1. They played boards they were not scheduled to 
>> play
>> and the result stands. ...  and, Sreve,  adjust the score by root(24/28) 
>> if
>> it's imps.
>
> Sorry, John, but I don't think that last is right.  If you are going to
> apply the 24-board VP scale, 24/28 is the right factoring.  It's an
> unbiased estimator of the score that would have been obtained in a
> 24-board match.  The square root comes in if you are establishing a new
> VP scale.  What this boils down to is whether you want to measure IMP
> difference (my view) or "sigmas difference" (John's view).  Either way,
> it's a very minor point.

I see your point. We can agree to disagree, but I have always thought direct 
factoring of imps is wrong (eg a 3/4 Howell butler pairs with a sitout) I 
want to play fewer boards if that's the case and get factored (because I 
expect to be above average). The spread of results, either in small fields, 
or with small number of boards is wider than large fields or long events. To 
correct for this we use all sorts of devices, and for imps I'm convinced we 
use sigmas.  Suppose they'd played 96 instead of 24?  Over half the time the 
result will be 20-0 won't it. cheers John
>
> _______________________________________________
> blml mailing list
> blml at amsterdamned.org
> http://www.amsterdamned.org/mailman/listinfo/blml