Re: Do I suck?
Some quick math assuming a 15BB/100 standard deviation.
Standard error of the mean=SD/sqrt(n)
Over 50000, we have 500 blocks of 100 hands, so n=500.
Standard error of the mean=15/sqrt(500)=0.6708
You can think of this as the standard deviation of your winrate. So if you want a 95% confidence interval, that's about two standard deviations each way. That means that after 50000 hands, you can be 95% sure that your true winrate is equal to your observed winrate +/- 1.34 BB/100. That's a pretty wide range. On top of that, 1 person out of 20 is outside that range after 50000 hands! 1/40 is more than 1.34BB/100 above their true winrate and 1/40 is more than 1.34BB/100 below their true winrate. In addition, to cut the uncertainty in half, you would have to play 4 times as many hands. Hopefully that helps clarify how big the variance really is in this game, and also why people typically suggest such large numbers of hands before you really start to put any faith in winrate numbers.
|