Winrate in small stakes limit games

[euroglot]

I played 1000 hands in the 0.25 - 0.50 limit games right know and I'm over 20$ down..

First I was in a huge winning streak and won like 60$ or so, but I lost it all and I lost 20$ total in 1000 hands..

I know there are huge swings in limit games but my question is:

After how many hand can you tell your true winrate?...When do you come in your long run?

[fishyak]

I play live and have very accurate data for last 6 months. A 95% confidence calculation for my results requires a range of 16BB/hr. to be statistically accurate. As you can see, statistically I cannot realistically "prove" I am a winning player.

You need 10k hands. And 10K hands may make you a better player. So even then, your true "winrate" may remain open to statistical question.

[euroglot]

hmmkaay...so you need 10K hands and then you know your 'true' winrate?...

[fishyak]

I "know" my winrate. But it is a range that is so large that I cannot tell if I am a long term winner or loser. So what I "know" isn't useful. Variance is so high and the winning margin is such a small number that you need a huge sample size to shrink the standard deviation calculation associated with variance to the point that 2 standard deviations is smaller 1BB/100.

But hopefully, if you play that many hands, you get better as you play and your sample may then overrepresent the impact of your earlier (worse) play. You see how difficult it is to get a statistically meaningful winrate?

In other words, don't pin too much weight on any winrate number. I see winrates as a benchmark, not a goal in and of themselves.

[destro]

The tradition answer would be 50K minimum sample size; 100K is better.

[Niediam]

[ QUOTE ]

After how many hand can you tell your true winrate?...When do you come in your long run?

[/ QUOTE ]

The answer is literally in the millions.

Plenty of examples around here where long term winners went on 50k+ hand losing break/even streaks.

[atnels]

It might not be a bad idea for the OP to google some of these terms in order to understand what a standard deviation is, etc. The math isn't that complicated and you can get a pretty good feel for everything using just Excel (or even a sheet of paper and a calculator).

Fishyak is saying that his stats provide him with something like 1.4 BB/100 with a standard deviation of say 1.2 BB/100 (I'm guessing). That means that 95% of the time, his true winrate will be somewhere between -1 BB/100 and 3.8 BB/100. Obviously this doesn't mean anything. But, if he had enough hands such that his standard deviation was 0.6 BB/100, he could be reasonably sure that he had a net positive winrate.

Make sense?