Quote:
Originally Posted by 2021shipit
I think if you are having 50-100k hand losing stretches at the micros you should probably reevaluate the games you are playing in and your own skill level. Doodoo doesn't really count since half the ACR winrate is rakeback anyway.
But if I put in a 5bb/100 winrate (which is quite conservative for 25/50nl on ignition), the probability of loss after 100k hands (using my friends STD, 90) is only 4%. A more standard winrate, 8bb/100, the probability of loss is 0.25%. At 50k hands, it's 2.3%.
Yes, but if you've played say 200k hands, that is actually 100k sub-samples of 100k hands (hands 1-100000, hands 2-100001,...., hands 100001-200000), so your calculation, whilst not being completely irrelevant, isn't anywhere near as relevant as you are making it out to be.
Incidentally, I don't think this mean that having 4% of your 100k sub-samples of 100k hands being breakeven or worse over 200k hands is "average" for a player winning at 5bb/100 with a 90 StDev (as those 100k samples aren't independent, only 2 of them are), just that the 4% is only applicable if you have played exactly 100k hands, not more.
The way I see this, if you have played exactly 100k hands and you are break even, then there is (according to your assumptions about winrate and StDev, and assuming static skill levels for both you and player pool) a 4% chance you are winning at 5bb/100. However, if a player has played over 100k hands and has a 100k hand stretch at break even (which is the information you are basing your calculation on), your calculation can't inform of the true winrate of the player.
I'm fairly sure this is correct, but please correct me if I'm wrong here.