No one will "know" their winrate, as observed results will always be the result of more or less statistical variance. The more hours you have though, the closer your observed results are likely to be to your statistical expected winrate.
Based on studies in this thread, even for those with lower standard deviation per hour, it's hard to be 95% confident that you're even a winning player with less than 500 hours. In other words, your margin of error plus or minus (which you figure with the formula below) is usually wider than observed winrate up until about then.
At about 1000 hours, the margin of error (at least for an average winning player) gets to be small enough that one can actually speak somewhat meaningfully about winrate, rather than just "I'm pretty sure I'm a winning player."
At 2000 hours you have a very nice statistical sample, but by then both the game and your play will likely have changed enough that the things that happened 2000 hours ago don't really tell you much about today's expectation.
Quote:
Originally Posted by HappyLuckBox
I propose from now on we should just post winrates based on a 95% confidence interval using the formula:
(2*stdvhrsbb)/sqrthrsplayed
where stdvhrsbb = standard deviation per hour divided by the Big Blind
and sqrthrsplayed = square root of hours played
So my standard deviation at 500nl is $343.93 per hour as shown in Poker Journal
And my hours played is 1333.23
My current winrate is $51.77 per hour
Plugging it in:
(2*(343.93/5))/sqrt(1333.23)
= 137.572/36.51342218965514
= +- 3.77BBs
So I am 95% confident my winrate at 500nl is between:
$51.77-(3.77*$5) and $51.77+(3.77*$5)
Or $32.92 to $70.62 per hour