True, it is only 21k hands so for sure it's subject to variance and unlikely I'm crushing for quite 13.2bb/100 if not impossible.
However one thing I think the Poker community don't realise is sample size is only one portion of what makes up what we can tell about someone's win-rate. Standard deviations and EV are just as important. A better mesure are confidence intervals.
So for my sample of 21k hands we can actually be more confident that I'm winning at a good win-rate than some samples of 50k or even 100k. Particularly as this is Ipoker where rake is high!