Accuracy of small samples (for opponents)
This may have been addressed at some point before, but I was curious about how well one could assess an opponent based on his stats and a small sample (200+ hands). I know that a small sample size will not yield anything accurate, leading to my next question...
because small variations in VPIP, for example, mean nothing in small samples, how large can I set the ranges for marking players? for a sample size of 400 hands, would 10% blocks work out? maybe 20%? Same goes for PFR, and hell, all the other GT++ overlay stats.
so far, the stats have provided me with a little extra evidence to probabilistically infer certain things. I just don't want to find out that the stat variations are VERY large after I lose 50BB bluffing and pumping on wrong reads.
|