Re: [Bug-gnubg] Measuring performance levels

[Joseph Heled]

Douglas Zare wrote:
> 
> Quoting Joseph Heled <address@hidden>:
> 
> > Douglas Zare wrote:
> > >
> > > [Snowie 1-ply]
> > >           Rollout      Money equity: 0.505
> > >                0.1%   3.6%  77.0%    23.0%   7.2%   0.0%
> > >                95% confidence interval:
> > >                   - money cubeless eq.: 0.505 ±0.013.
> >
> > at 0ply, 12960 games I got
> >
> >   0.07% 3.6% 77.06% 23% 7.08% 0.0%
> >
> > So this is the same as the above (SN 3?). I leave higher plies to someone
> > with a
> > stronger machine.
> 
> It looks the same, but since Snowie does not use variance reduction for 1-ply
> rollouts, it might be that Snowie plays slightly better or slightly worse.

I did not use variance reduction either. I still don't understand why it works,
and was not able to prove to myself I get better results using it (as opposed to
doing more trials).

I have the (individual) standard deviation for the rollout, but can they be
converted to a confidence interval?  The numbers are (0.391 0.133 0.022 0.219
0.001)

 -Joseph