Oh I know if I'm a winning, losing, or break even player.
I simply took out my win rate because it doesn't actually matter.
I just wondered if there was a thread detailing confidence intervals for offering numbers of hours or anything like that.
After 500 hours, if your win rate is 2BB, there is a 95% chance your true win rate is 2BB+/-3BB. Stuff like that.

Your variance is going to change depending on your play style.

Given that there are so many human factors unrelated to the math itself, such calculation would be rather unreliable.

Why is it important to figure out WR anyway?

I'm concerned as to why you think winrate is not important?

(over a sample size of 1k+ hours)

It is important, but without clear intent or purpose, such number has no meaning.

Do you intend to study harder if your win rate is 2bb rather than 6bb, or do you just stop trying to improve your game if you hit 8bb/hr?

Do you move up stake if you hit 8bb in your level?

I mean, what real purpose is there to determine confidence level of your WR?

Winrate really isn't that important. Your game is always changing. Over 1k hours your game should have evolved. You are never really going to know your true winrate. You just have a good idea of how you are doing in the game.

I would imagine a winrate around 700 hours would be accurate to within +/- 1bb

It can be important when used with standard deviation. One can project long-term winnings (or loses) with relative accuracy.

lol 700 hours is like 15,000 to 22,000 hands, thats enough to determine if you are a winner or a loser but not to even remotely know what your theoretical expected value is in the game.

I highly doubt that's the case. In the grand scheme of things 700 hours really is a short sample.

winrate is just a number that makes people feel warm and fuzzy or ****ing puke. It either makes them think they are good or bad. I remember when my winrate was sky high and I was like "LOL LIVE POKERZ" then as it has slowly decreased over this extended break even stretch (or huge ****ing downswing depending on how you measure it) it isn't such a nice and fuzzy feeling any more. Regardless, it isn't like a day job where you show up and get paid (x $/hr) and all you can do is play as good as you can and see what happens.

I mean I can probably find thousands upon thousands of dollars that were left on the table through missed value or bad calls or bad folds but there isn't much that can be done about that now. All you can do is play better. However optimal play, much like hindsight, is 20/20.

Same general feeling to what 11t said.

If anything, I feel less inclined to work on my game when I am running like a god.

Maybe this is just me, but I can take any chunk of 500 hours from my session logs and my winrate always falls between 12-14bb/hr, so I just go under the assumption that I'm going to make ~13bb/hr. Do you guys find similar patterns?

I've only been playing poker as my sole source of income for a year now, but using that 13bb/hr as a model has allowed me to accurately pay my bills in advance and project profits.

Obviously it's not completely accurate, but it is quite useful as long as you compensate for a reasonable margin of error.

Using poker WR to calculate future earning is very optimistic thinking.

There are way too many factors involved, such as changes in poker room traffic, new sharks, dead whales, and so forth.

Plus higher your WR, lesser the likelihood of sustaining it.

Eventually there will be a 500 hour stretch with different results. If all of your sessions had almost identical results, only then would it be possible to have that consistency forever. I don't doubt you 1 bit - so don't take it the wrong way. It is just probability and statistics that eventually the x% chance will happen where so many bad sessions (or good sessions) happen consecutively.

Kinda doubt it myself. My massive run good lately has destroyed any confidence I have in my winrate. At my lowest in 2012 my overall was running at about ~$21/hr and now my overall is running at almost ~$30/hr, just a mere few hundred hours later (a 3bb/hr swing).

Not even close. Go back in the thread, there was an estimator put out you can plug into excel to estimate your confidence. At normal SD 700 hours is enough to give you a rough idea of where you lie on the losing, breakeven, winning, crushing scale, but that's it. If your win rate is 12-14bb/hr, that's a crushing rate, it's about 98% that you are at least a solid winner (5-6bb/hr minimum) if you maintain that for 700 hours. The chance that you are doing at least 10bb/hr (a reasonable crush threshold) is more like 85%. The chance that you are doing at least 12bb/hr (your +/- 1bb/hr) is more like 60%.

Your 95% confidence interval is something like +/- 6 bb/hr at 700 hours.

And that all assumes a normal distribution and no prior information which are both suspect/bad assumptions.

For one thing, any winrate above 10bb/hr is pretty unlikely, it requires both excellent play *and* good game selection. So my prior is that any rate above 10bb/hr involves a much greater likelihood of run good than of run bad.

no way your winrate is +/- 1bb after 700hrs... I've had my winrate go +/- 1bb after sessions

Boring question but i was wondering how many losing months you regs have had in 2012?

say guys that play 120+ hrs avg month

Im starting to put more hours in live games and trying to read as much as possible before i dive in again

Thanks

Yep, LOL at 20k hands being a significant sample size.

I had one losing month but that's because I went on a 2k+ downswing and took the rest of the month off. I probably don't average 120/hr month though.

0

I had a brutal 8k downswing in 2/5 last year and still managed to have a winning month. Actually, I've never had a losing month.

I'm curious how much this estimator was based on online vs live. I would imagine playing live would result in higher confidence because we don't have as large of downswings due to the vast disparity in talent. I want to go back and see it, but there's a ton of posts in here that I would have to wade through. Or is this a common equation in statistics that can be found online and is completely independent of poker?

I'll concede that I'm probably wrong in guessing that one has a 1bb margin of error at 700 hours, but in your opinion (and others), how many hours would it take to get a margin of error that low? Is it even possible? What would an average standard deviation look like? After just over 1500 logged hours, my SD is at about 80bb/hr and a 13bb/hr winrate. How does my SD relate to most other 10+bb/hr winners?

I hit some rough patches late last year, but there was a period of roughly 600 hours between March and August that was ~10bb, and my standard deviation was 51bb/hr, session was little over twice of that at 107bb/hr.

Given as consistent as my result was for what seems like an eternity live, there was no consistency when looking at the overall picture.

Because poker is a zero sum game, and I do not reside in one of the poker paradises, my competition is always improving by weeding out the weak and forcing the remaining to adapt.

So ya...there is no consistency in poker. We are humans after all.

Unless you're playing rake free poker, it's not zero-sum.

Well that's the hard thing to figure out about live poker. What exactly is an accurate sample size. The thing with 700 hours (21K) hands is that is it is much more accurate than 21k hands of online poker since winrates are bigger, but its not accurate enough to make any meaningful conclusions.