Using Small Samples Cautiously (mathy)

[Sigurd]

Thanks for the search phrases, I'll have a go at it later.

I were wondering if it would be possible to show how the reliability increases with the number of observed hands graphically. Personally I have a much easier time understanding math if I can see it represented graphically.

btw. nice post.

[Christian_Peters]

I'm on break from school right now, and I don't have access to statistical softwear (other than PT [img]/images/graemlins/wink.gif[/img], but any stat work I do in the future, I'll try to include graphs from probably STATA or SPSS.

[Aaron W.]

[ QUOTE ]
I'm looking for ways to corroborate stats based on small samples with other stats. This is one reason why I keep people's BB/100 winrate in my PAHUD popups. Yes it's meaningless by itself, but if I see someone with 74/50 stats after 25 hands, I look at the winrate. If it's 250 BB/100, they've simply had a hot run for a few orbits. If it's -98 BB/100, he's most likely a donk.

[/ QUOTE ]

I think think the bolded inferences are very suspect. Winrate has such a low correlation to ability over small samples that you may get something like a .1% shift in confidence. The phrase "most likely" feels like a strong overstatement.

[fabadam]

[ QUOTE ]
[ QUOTE ]
I'm looking for ways to corroborate stats based on small samples with other stats. This is one reason why I keep people's BB/100 winrate in my PAHUD popups. Yes it's meaningless by itself, but if I see someone with 74/50 stats after 25 hands, I look at the winrate. If it's 250 BB/100, they've simply had a hot run for a few orbits. If it's -98 BB/100, he's most likely a donk.

[/ QUOTE ]

I think think the bolded inferences are very suspect. Winrate has such a low correlation to ability over small samples that you may get something like a .1% shift in confidence. The phrase "most likely" feels like a strong overstatement.

[/ QUOTE ]

I'll fully agree that 2 hands shown down are probably more of a read than the above. I use this stat mostly to caution my conclusions: if I see 74/50 stats and a high winrate over a small sample , I'll asume a hot run rather than donkishness. The reverse is harder (that is, cold run vs donkishness), but it can help to confirm my read.