[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: [Bug-gnubg] can someone explain this please?
From: |
Joern Thyssen |
Subject: |
Re: [Bug-gnubg] can someone explain this please? |
Date: |
Mon, 21 Jul 2003 10:55:15 +0000 |
User-agent: |
Mutt/1.4.1i |
On Sun, Jul 20, 2003 at 02:57:09PM +1200, Joseph Heled wrote
> Error rate (total) -0.015 ( -0.046%) -0.117 ( -0.356%)
> Error rate (per decision) -0.001 ( -0.002% ) -0.007 ( -0.021% )
> Equivalent Snowie error rate -0.4 -3.4
> Overall rating Supernatural Expert
> Actual result 52.94% 47.06%
> Luck adjusted result 49.47% 50.53%
> Relative FIBS rating -3.68 3.68
>
>
> X came out Supernatural vs. Expert, but his luck adjested result is 49.47%
>
> What's the catch?
Based on the error rate X is 50% - 0.046% + 0.356% = 50.3% favourite,
but the luck adjusted result is 49.5%.
I guess this is the kind of differences you expect to see between biased
and non-biased results. If X played gnubg-like the "overall" rating will
be biased towards him.
Also, the luck adjusted result come from a 0-ply analysis whereas the
error rate comes from a 2-ply analysis hence you'd expect to see
differences.
Since this surely will be brought up by others I might as well write
a section for the manual that I can feed to them.
Jørn