Well, it really depends; some stats are obv useless unless you have a decent sample but I think ur missing out on some key issues:
first of all, the fact that u need a large sample to have a high confidence interval is true but that doesn't mean that little info = no info.
Even if we have small samples, those the stats are more likely to give us a basic idea - than they are to throw us completely off.
It's actually a complex issue but I don't wanna go into it in depth unless someone is interested in this issue specifically
.
In your post you mentioned Vpip; Vpip is actually quite reliable very quickly.
Let's play a little game:; one orbit, 8 hands; Villain has a vpip of 75%
How likely is it that villain is actually a 18/15 guy?
Answer? about 1 in 1800
That's pretty decent, wouldn't you say?
let's try a more realistic example:
40 hands, villain has 32 vpip; what are the odds of him being the same 18/15 guy?
Answer: about 1 in 50?
not too shabby, right?
I'd put my money on 49/50 wouldn't u?
obv, if you wanna know if villain has 22% or 24% you'd need a bigger sample
But my point is- some stats start becoming reliable much faster than you think, especially if it's a rough estimate ur after.
P.s - a little clarification, I did the calcs here pretty much instantly and what they actually mean is this: Villain A will actually be "looser" than 18 Vpip 1799 out of 1800
times.
Meaning he will have 18 (or less) once in 1800 times
but off course, he can be 25 for instance. In fact he will be 25 (or less) about 1 in 250 times
I hope u get the idea.
This was one messed up post, srry guys
.