2 sigma certain 31.51 is the standard deviation of his "hourly result" from each session.

With thanks to bip! math, updated std dev for last 7 months of results:

Most recent 440 hours
Cashed 37/54 for 68%
BB/hour 10.69
St dev 84.97 bb/hour

Most recent 329
29/41 for 70%
BB/hour 12.73
Std dev - to much of a pain to do on android spreadsheet app

Most recent 209 hours
19/24 for 79%
BB/hour 17.27
Std dev - to much of a pain to do on android spreadsheet app


So; fish on a heater, getting better or opponents getting worse?

Or completely meaninglessness?



Sent from my SM-T320 using 2+2 Forums

Unexploitable in the sense that opponents cant adjust their game in any way to exploit you.

If you dont believe the above to be a true statement about "GTO", then its you who dont know what GTO means.

Zippy - very nice results.

Breaking it down to sub-periods & trends is akin to finding faces in clouds... look hard enough and you will see whatever you want to see.

Best answer

I do not like the attitude prevalent in this forum that such a sample size is basically meaningless. You can quantify the "meaningfulness" and decide for yourself what you think of it.

To give an idea, with that observed SD and BB/h over a 440 hour sample in theory your true winrate is 95% to fall within the range (10.69-1.96*84.97/sqrt(440), 10.69+1.96*84.97/sqrt(440)) = (2.75 BB/h, 18.63 BB/h)

This page has info for calculating confidence intervals for a normally distributed data set https://en.wikipedia.org/wiki/Standard_error

In reality this is not perfect because results are not normally distributed over a 440 hour sample AFAIK. Also, we know based on many others' results that the upper part of that range is not a sustainable winrate.

You can see that such a sample size only tells you that you are likely at least a small winner. It's still reasonably possible, but unlikely, that you're really a fish on a massive heater.

My logs show a similar trend as yours. My most recent 200 hours has me at 23.0BB/h while my entire 456 hour sample has me at 14.8 BB/h. I know I am improving but the sample size isn't large enough to demonstrate that fact with reasonable confidence, and 23 BB/h would likely be unsustainable even for an elite super-crusher

You may not like it, but the truth is it is largely meaningless from a purely statistical standpoint. Unless you've played a a few hundred thousand handsonline you really can'thave perspective of what variance can do to you over a few thousand hands. 440hrs is at max 13k hands. The higher the win rate, the more likely you're a winner of course, but you can't be certain just based on the results.

We can also judge based on how you play and win. At this point, we have a pretty good idea about what winning players do. So, if your game looks a lot like many many other winner's game, then you're probably a winner. If you have a 25bb/hr winrate over several hundred hours and you're posting insane HH's and sh*t strat, I might start to wonder if you've been on the good side of variance.

My objection is only that people like to describe samples with a statistically quantifiable indication of reality as meaningless. They just say it means nothing without any qualifiers.

Whether or not a sample is meaningful to us depends on what question are we trying to answer. If we're comparing win-rates among players of similar ability then a 440 hour sample is not nearly enough. If we simply want some validation that we're beating a game then a 440 hour sample is very meaningful, especially in conjunction with other players' evaluations of our ability based on our HH's and strat postings.

Couple this with the arguments often made in this thread that in live the high likelihood that you are a different player and\or your specific poker environment has changed over X hundred hour period; all win rate discussion are at best an interesting thought exercise.

And so the cycle of life is repeated

I have several hundred thousands hands of online play in my poker tracker db, but latest entry is from April 14th a bunch of years ago. Stopped playing after that for a long while.

Live low stakes is nothing like grinding 50nl online. Not even close. 1/2 players are for the most part just awful. Bet/fold for value, calling poker is loosing poker, play an exploitable style for decent lower variance income from your hobby.

How often do you sit at a 10 handed 1/2 table and after 60 minutes feel 100% confident that 6 or 7 of your opponents have ZERO chance to not go broke overtime unless they stop playing? Only variable is how much time and what maximum win rate is theoretically possible with time being the 60 minutes per hour constant in the equation.


Sent from my SM-T320 using 2+2 Forums

Correct, 440 hours are probably sufficient t know whether someone is a winner.

But most people who post these charts aren't asking whether they're winner, but rather how big of winner they are. If you have won 3000bb after 300 hours, do you really need others to validate that you win?

Agree. I just find theoretical discussion of what maximum rate is possible interesting.



Sent from my SM-T320 using 2+2 Forums

LLSNL and 10nl/5nl/2nl actually compare quite similarly, especially when you head over to the micro forum or beginners forum and see all the people who can't beat 5nl you start to see why some many people can't beat 1/2 live.

The best though is when your playing live and terrible Vs start talking about how they play online at the exact same level they play live I just wish I could find these guys to play 200nl/500nl online for 80-120 hands/hour

I remember reading an article in Card Player magazine, where the author of the article pitted the same Hero Bot vs. the same line up of Bots & played enough hands for a lifetime. He did this 4-5 times.

In one instance, the Hero was subjected to a losing streak that the author did not believe was possible. In another instance, the Hero never had a meaningful losing streak.

I don't remember if it was 1,2 or 3 million hands per run. 1 million hands would take 20 years @ 25 hands per hour playing 2000 hours per year.

So, although the results over 7 months, totaling 790 hours are meaningful to a degree, I choose to disagree that it is anywhere near what is required to be considered solid evidence of a player's skill set.

I've logged 1421 hours since I started using a session logger app in Feb2015 & I still can't decide how much weight to put on my WR as evidence of my skill set.

Anyone play with a rake similar to 5+2 ?

I know I'm getting slightly higher rake then most places but im curious as to amount if rakeback people are getting. Right now I'm sitting at 2500$ in promotional money for 2016 which I know is insanely high for only 180 hour played. Just curious if anyone else has stats I can use to compare.

By the time you have any meaningful data you should have improved to a point were your data should not reflect your current playing ability.

Yeah, I remember the table fish at my 5/5 game saying he played 5:5 and 5:10 online and just being aghast. I actually felt bad for him as he talked about how rigged it was and how he could never win.

Buy him a drink and agree that yes, the $5/10 on PS is rigged.

are you counting comp dollars?

I think we all generally agree on this. The statistics although telling are not concrete evidence of winrate or even winning at all. We have mountains of anecdotal evidence however that can be used to justify the winning (or losing). I don't have near enough data to show that I'm a winner or how much I'm winning. I am 100% certain though that I am one of if not the best player in my local pool. Whether results bear that out is why the fish keep coming back (I might be the fish).

Are you taking about things like high hand prizes?

He must be.

What kind of equity are you getting in those promotions? You have basically been getting $14/hr in promotions. If that's on par with the equity, then I wouldn't have any problem with that drop. But if you are getting $2/hr in equity, then you are obviously running way hot in variance and I wouldn't exactly be too excited with that structure.

Keep in mind, most promotions do bring in players and increase overall equity for winning players because of various factors.

Just about everywhere in FL is some variant of 5+2, also MDL.

My very rough guess of tyipcal high hand promotional equity (rakeback) is $3-$5 / hr. But it depends, of course. Lots of variance. Depends when you play, what you play. yada, yada, yada.

Since the Bots' skill set does not improve over the course of 1 million hands, the author's test did nothing more than identify how much of an impact that variance, may or may not have, on a player's results over time.

I play ~110 hours per month, so over 6 months that's 660 hours. If my true WR is Xbb's pr hr & my skill set does not dramatically decay or improve, I am certain that my results over the next 660 hours could be off considerably, one way or another, from my capability based upon my current skill set.

Furthermore, there are numerous variables a player encounters playing live vs. the author pitting a Bot against the same 9 Bots.

So why do I track my WR? For the sole purpose of identifying my profitability.

Identifying one's change in how he plays is more crucial to me. I keep a record of all my bad & good plays and strive to limit the percentage of bad plays.

A few days ago a wild player, with whom I've played against for over a year, sat down at our table, as he pulled ~$200 out of his pockets. He won a hand or two & then got involved with me HU after the flop. I bet the turn with AK on a KQxx board and he snap shipped ~$145 which was maybe $35 more than the pot. As I pondered whether to call, I gathered chips while observing him. He chewing his gum as if it was his ex-wife that he was preparing to spit out & I finally deduced it was at least 50/50 [if not in my favor] as to whether he had me beat & I called.

I won, that time, and I was suddenly up well over $400 in a 1/2 game that I'd only been playing in for ~2.5 hours, as I had been winning before that hand. It gave a nice little boost to my 2016 WR, which totals 190.74 hours already.

You don't add rakeback or promotion money you won, on to your win rate. Your win rate is the amount of BBs per hour that you win from skill, not luck or gratitude from the casino for playing there.

here we go again....

It's good to know both figures but you def can't ignore it. If the room didn't rake for promos at all your win rate would be higher, so why contribute to the fund but never factor it in when you get some back?

Obviously you don't want to plan your future based on a $50/hour win rate at 1/2 because you ran super hot in promos recently, but I think it's a pretty easy situation to use proper judgment.