Something has been bugging me lately with stats.

In a game with 6+ players obviously a larger sample size has huge benefits mainly due to the fact that overall most players will act pretty much the same with no need to change overall.

However in a heads up game people are constantly trying to adapt, change and mix up their play to get an edge on their current opponent and how they are behaving.With this being the case, to me it seems that the larger the sample size of an opponent in a heads up game the more meaningless/deceiving it would be.

For example if your opponent notices your cbetting 80% of flops and he is folding to 70% of them he would most likely start floating a lot or 3betting to counter the high amount of flops you cbet. Because this is a sudden change to his gameplay, if you have say a 100 hand sample on them the previous high fold rate is going to take a long time to display/converge their now high 3bet/float rate because of the sample size. Without previous hand history this would become obvious in 10-20 hands.

While a change like this is going to be obvious to a good player, in my opinion it's always great to have good stats to backup your thoughts and influence your decisions.

Of course I do have a question to go with all of this. If your using stats in HU play or HU SnG play what sample size do you use and/or think is optimal? max history? Past 50 hands? Past 10 hands? or other?