Monday, March 10, 2008

Luck

My bankroll cracked $1k on Stars this weekend, so barring a downswing I'm well on my way to meeting my March goals.

That's beside the point for today though. As I'm sure you've seen in this blog, I always like to review my sessions to evaluate my play based on my Sklansky bucks versus actual winnings user PokerEV. It's a great tool to "remove" variance from analysis, but unfortunately it really only equalizes people drawing out. You could still have a session of playing "perfect" poker, run your KK into AA five times, and drop several buyins while having Sklansky tell you that your EV was horrible (because it was, against the villians actual hand, but likely not against his range). Thus, while Sklansky buck evaluation can give you a "better" idea of your true winrate, how many hands do you have to play before your heater/cooler hands inevitably equalize?

Obviously someone on the 2+2 forums with a background in statistical analysis was asking himself the same question, because they posted this thread. If you've ever wondered just how significant variance is, I definitely recommend reading it, even if you don't understand the math.

The result is clear: variance is huge, and until you've played a very large number of hands, you should most definitely not be underestimating luck in poker. For comparison, I'll have to log ~125,000 hands total to be able to say with 95% confidence that my true winrate is within 2.0 ptBB of my actual. 2.0 ptBB is a huge margin, and to reduce that further will require a significant additional number of hands (roughly ~1 million to reduce the margin to +- 0.5 ptBB). As it stands now, over ~45k hands I'm a ~3.5 pt/BB winning player. According to Sklansky bucks I'm running cold, and I definitely don't feel like I've had any more heater hands than I've had coolers. All that said though, it would be perfectly reasonable for my true winrate to be significantly lower (though I maintain it's probably higher!) than my actual, based purely on the math.

No comments: