bb/100 question

I think the standard for most people is around 100k hands for an accurate guess.

After 25k hands you have a pretty good idea.

Accurate you should have a Million hands for, but 100k gives you a good hint.

you can't ever be positive that your winrate is pretty accurate because poker is too dynamic.

okay, that smartass answer was uncalled for. but jbpatzer wrote a pretty interesting post on the subject

Originally posted by jbpatzer
After all these posts on variance, I thought you'd probably appreciate a bit of maths. I wrote a chapter on this for me and TT's book, so I've given it some thought.

At NLHE, a decent win rate is 5bb/100 hands and a typical standard deviation is 100bb/100 hands. In 100 hundred hands you can easily stack someone or get stacked once or twice, so, although your expected winnings are 5bb, the random nature of the game means that this is plus or minus about a stack, hence 100 hands is absolutely bugger all. We all know this, but how many hands do you need before it isn't bugger all. Winnings grow in proportion to the number of hands, but standard deviation grows in proportion to the square root of the number of hands. Look at this high tech, professionally produced table.

Hands | Expected Stacks Won | Standard Deviation
------------------------------------------------------------------
0.1K | 0.05 | 1
1K | 0.5 | 3.16
10K | 5 | 10
100K | 50 | 31.6
1M | 500 | 100
10M | 5k | 316

100K hands is bugger all, and 1M isn't much more than bugger all. Tbh, you probably need about 10M hands, so unless you're leatherass, you're basically fucked. I hope that's cheered everyone up.

Mathematically speaking it's a lil' hard to say, but considering you play at least 500 hands/hour you need at least 100k to have an idea, after that you will just be getting most precision on the results.

By the time you have an accurate sample size, you would of changed your game, so it's never going to be accurate, unless you play some robot style for like 100k hands +

Its next to impossible to say what your actual winrate is, only what your winrate was for the last X amount of hands.

You should check out variance simulator at http://www.evplusplus.com/poker_tools/variance_simulator/

If you play around with it you find that even on a 250k hand sample size a 5bb winner can either run at 10bb if his lucky or is breakeven if his unlucky.

And even on a lets say 5 million hand sample your winrate isnt accurate because you have likely improved your game in the process and even on that sample size the winrate can easily be 1bb off.