So I just very recently started using PokerTracker for Omaha and have logged about 22,000 hands at 25/50. Aside from specific player win/losses though, I find that I don't really use the program for anything. I talked to my friend (who is a NLHE only player) and was told that 5-7 BB/100 means that you are a solid, successful player, and that anything over 10 just pretty much means that you are playing at too low of limits. Does this hold true for Omaha also? Thus far, over about 22,000 hands, I have a 14.20 BB/100 ratio. Should I stick to the current limits or move up? Thanks for your help.

Also, are there any other statistics on PT that I should be using to analyze my and everyone else's play..

at the lower limits, a good PLO player will have a higher winrate (in PTBB) than the same NLHE player, simply due to the fact that most people are at least competent at NLHE these days.

I can't say re: the super high stakes, I would imagine that most of your competition is more than competent, but I do know that 22k hands is far to small a sample to determine your true winrate in those games.

I figured as much but my NLHE buddy told me that 5k was usually enough.. So he is just flat out wrong then?

although i've taken several stats courses in both undergrad and grad school, i (maybe purposely) didn't digest too much of it -- Pete, Troll, and a few others are the real stat guys that can tell you when the sample size will start converging towards your long term winrate, but it isn't 22k hands and most definitely isn't 5k, particularly in a game like PLO where your SD is significantly higher than NLHE on average.

but one thing i can say: move up when you are running good and feeling confident. if you take a hit, move back down. this advice can certainly apply to all stakes, even nosebleeds.

That's a pretty hot winrate. 22K hands is nowhere near enough to gauge your true winrate accurately, unfortunately, and you could just be running very hot. In any case, it's unlikely to be deviating too far from your true winrate, and you're likely a big winner in your game. If you're rolled, I say move up.

Thanks guys. So what is a reasonable number of hands to judge longterm winrates then?

i'm too lazy to do this right and come up with a general formula for you, but i've calculated the standard deviations on winrates over a number of hand milestones (using my sd/100 of 90):

10k hands: 9
20k hands: 6.36
50k hands: 4.02
100k hands: 2.84
200k hands: 2.01
500k hands: 1.27
1m hands: .9

so a 95% confidence interval should be around +/- 2 * the number above. e.g. about 5% of 6bb/100 winners will run at 8bb/100 over a million hands. and just for you, gordo, the sd over 22k hands is 6.07 bb/100. so in theory, a 95% confidence interval would be +/- ~12bb/100, so there's still a decent chance that your "true" winrate is below 2bb/100. specifically, about 5% of 2bb/100 winners will have your winrate over a 22k hand sample.

while this is a good place to start, it doesn't actually create a confidence interval for *your* winrate, since when you have a high observed winrate it's much more likely that you're running good than that you're running bad (e.g., there's really no chance that your true winrate is 26 bb/100). this could be corrected for in theory, but you'd need a good model for the distribution of winrates over the general population of players and I have no idea about that.

I think i made a mistake in that last post. At the 95% level, it should say that 2.5% of the people with a true winrate of 2 sd below your winrate will run as well as you over whatever # of hands, since i used the z-score of a 2-tail distribution.

Anyway, I've corrected that, and I also wanted to make something people could reference in the future, since this question is asked like every other day. So here's a generic table based on an SD or 80 and wr/100 of 8. The "x% min" columns mean that that %age of the people with the winrate in that column will run at 8bb/100 over that number of hands. E.g., over 20k hands, about a third of people with true winrates of around 5.7 bb/100 will run at or better than 8bb/100, and about 5% of people who are 1.3 bb/100 losers will run at 8bb/100 as well.



Note that these outcomes are dependant on the standard deviation but NOT on the winrate. So it's perfectly fine to plug in a different winrate, and just take the winrate and subtract the z-score for the %age you want times the number under "sd in bb/100." The z-score for 33% is .439, for 25% is .674, and for 5% is 1.645.

also something to bear in mind: people are much more likely to come here and say "is this a good winrate?" if they have been running hot. Really, you should measure your winrate over the next 50k hands or whatever, not the last 50k.

yes, and again the table i posted can't tell you what your winrate actually *is*, only how probable or improbable your run is for a given winrate.

All good stuff in this thread.

But I think the basic answer is that it's pretty much impossible to use older data from your poker database to predict what your true winrate is mostly as it relates to looking forward.

Time is much better spent with Pokertracker and PokerEV analyzing your play and making sure that you are getting your money into the big pots as a favorite and that you don't have any leaks in smaller and midsized pots.

I understand this sentiment, but it's not strictly true. Past winrate obviously correlates to your "true" or "future" winrate pretty strongly, and more and more so as your number of hands played increases.

One interesting way of looking at this is that the common view on what NLHE guys make is 3-4 PTBB/100, at least once they get to semi-sentient limits. SO if you are making more than this, it was better you played PLO than NLHE at least.

gl

bdd

Good thread.

I have acouple of question. First, pete:
If I supposedly have run at 28 BB/100 for 5k hands, what are my likely winrate-interval based on your table? I didn't quite get your explanation.

Second, what is PokerEv, and is it useful?

can anyone calculate or estimate the standard deviation after adjusting the winnings with the pokerev program? standard deviation should go down a lot and you should be a lot more precise in you estimated winrate.

I don't think the adjusted winnings in the PokerEV luck graph actually tends to your true winrate (I think Pete said something like this too recently). According to Phil the author, doing the calculations the more natural way (you get lucky or unlucky for the whole pot) is also subject to bias, but I don't see how.

Just to make to make you notice that the standard deviation/100 given by PT is false. PT calculate the SD related to a session and not to 100 hands and divide by the number of the session to give the SD which is not at all correct. I sorry if you all not this but I thought it might help. This subject has been greatly discussed in the SSNL forum and I think was first discovered by Pokey.

Is this a recent bug discovery, or a change in the way PT calculates SD? I'm pretty sure Pokertracker Pat posted to say it was calculated on a hand-by-hand basis.

old discovery so maybe it has been corrected since but I haven't noticed any change in my SD (in NL) so...

the result has to be better in general if we consider if we got lucky in all in situations or not. there is just no point in considering yourself as a like 5bb winner if you just got lucky for that amount in flips. the question is just how much it reduces our variance.
statistically there are some problems with the whole winrate calculations because you need to make some assumptions:
1. table conditions are constant over the time
2. no autocorrelation between our sessions (this might not be true as a lot of people will go on tilt when losing)
3. there is no improvement/decline in our game over the time.

i probably forgot some other assumptions.

here is the poker tracker thread about it

http://www.pokertracker.com/forum/viewtopic.php?t=16455&highlight=standard+devia tion

it really shouldn't make much of a difference if you have a large sample.

LOL, I was going to say that I remember asking that same question. Then I click the link and find out that you linked my thread. Deja vu all over again.

the chart doesn't purport to give you an "interval" of any kind, it just tells you how frequently you should expect your hot-streak to occur for players at various winrates. i will tell you that a break-even player who plays 5000 hands will go on a 28bb/100 heater about 1.5% of the time, where a 10bb/100 winner will go on the same hot streak about 8% of the time.