Re: [Bug-gnubg] Snowie Equiv error rate

[Joern Thyssen]

On Mon, Jan 12, 2004 at 04:36:11PM -0500, address@hidden wrote
> Hi
> I analysed a match with Gnubg and S4.
> Gnubg's "Snowie equiv error rate" doesn't match S4 error rate.

What do you mean by doesn't match? Can you provide an SGF file as well
as the detailed statistics from Snowie, please?

The "Snowie equiv. error rate" is calculated with the (almost) same
formulae as Snowie, but it's obviously gnubg that calculated the equity
loss. 

You shouldn't expect them to be completely identical. As I don't have
Snowie I've never investigated what deviations you might expect. +/- 1
to 2 is not unrealistic given that gnubg typically rates snowie with
error rate around 1-2 and vice versa.

> Also can you show this figure to 3 decimal places please.

You control the number of figures in the output from
Settings->Options->Display->Number of digits. We subtract 3 from the
number given in the settings, as the error rate is multiplied by 1,000
before it's displayed.

Jørn

pgpbrqz2LCNfw.pgp
Description: PGP signature