Victor,

I could ask people to break down their winrates by position, i.e.

Button #times, winrate
Cutoff #times, winrate
SB #times, winrate
BB #times, winrate

We would need about 4 times the number of people responding to THAT request to get a statistically sufficient sample (actually many more than that since there will be fewer samples in the earlier positions).

Even more, we can ask people to submit their VPIPs in each position along with their positional winrates. But then, we run into some problems: what does it mean if one guy plays K9o a lot more in the big blind than another guy but his winrate is lower? Does it mean he should tighten up? Or could he just be running bad? To answer these questions we would need twice as many people as before, one group with higher vpips and one group with lower vpips. So, to even get close to a statistically sufficient sample size, we need to ask for an all-inclusive stat (like overall winrate) that people will have a lot of samples for.

The other major reason to stick to simple overall winrate is that if you ask for something complicated, you will get even FEWER people responding to your post. If you don't believe me look here:

http://forumserver.twoplustwo.com/sh...age=1&vc=1

-v