Re: [Bug-gnubg] Re: The importance of METs

[Nis Jorgensen]

Joseph Heled wrote:
> Another run,
> 
> E1 - 31747
> E2 - 31994
> E3 - 186259
> 
> Total -
> 
> E1 - 63494
> E2 - 64061
> E3 - 372445
> 
> Does this help?

AS usually, the confidence intervals are inverse proportional to the
square root of the number of trials. Recalculating the mean, I get a new
95% confidence interval of

50.11% +- 0.07%

So we are starting to be confident about the results (although not very)

I am academically interested to know whether luck reduction could
improve on this (I am quite sure it is not worth the time). Could you
try doing it on a sample of perhaps 1000 matches, and send me the
results (raw results are fine, as long as I can identify matching matches)

BTW: You seem to be sending you replies to me only. Is this intended?

Nis