Quote:
Originally Posted by LordRiverRat
I'm reading some Ed Miller right now and he said at 2/5 if you're a winning but not elite player "would'nt shock him" to hear about you going on a $30k downswing. That's 6000 bbs playing live poker. WTF?
Well...
The math:
A winning player might have a std dev of 100BB/hour (perhaps more - I think mine is currently calculated at 120BB/hour). The standard deviation per N hours grows with the square root of N. So this would be the same as a std dev of 1000BB/(100 hours) or 2000BB/(400 hours).
However, 2 or even 3 standard deviations shouldn't be "shocking." (3 standard deviations below expectation is about 1 in a 1000). So a winning, but very unlucky, player could find themselves running 6000BBs below expectation after 400 hours, if they're playing some fixed strategy with a std dev of 100BBs/hour.
Add in not too many more hours at a lowish winrate, and you've got a 6000BB downswing over hundreds of hours.
The reality:
I tend to believe that there's something amiss in all of the straight stat stuff. For one, I think it's possible to play a winning style with very low variance - much lower than 100BBs/hour - and I think most winning players would naturally gravitate towards this style when on a bad streak (and probably just quit before going 30k in the hole). And I think a lot of what "variance" actually is in poker is when players occasionally make very bad decisions, then correct for those bad decisions in future play. For example, maybe a winning player tilts off a buy-in once every 50 hours, but not more frequently. Although this would affect the variance calculation, it's more of a constant effect on the winrate than actual statistical variance.
tl;dr:
Yeah the math seems to imply that a 6000BB downswing is very possible. But the math might actually be wrong in terms of what actually happens in real life.