Strange Statistical Request (Sanity Check)
Posted by HawksWin
Posted by
HawksWin
posted in
Low Stakes
Strange Statistical Request (Sanity Check)
Hello Gentlemen, I was playing around with some stats the other night and came across something I found to be alarming. I have been over this over and over again in my head and I am quite sure I am not missing something.
The PT4 database on my laptop contains 862,840 hands. All of these hands have been played on Ignition.
So, I got to thinking the other night and I ran a filter to see how often I was being dealt the top 5%, top 10%, top 20%, etc. of hands. Based off of my sample, archaic math (no variance involved) would dictate I should have been dealt a top 5% hand 862,840 * 5% = 43,142 times. When I run the 5% filter in PT4, I was dealt a top 5% hand 40,199 times. Obviously, these numbers are going to be different, but a spread of 2,943 occurrences or 6.8% seems very high (extreme run bad) for a relatively substantial sample. In the grand scheme of things, close to 1,000,000 is a pretty small sample with the exponential nature of numbers.
So, my request, is that those of you with decently sized databases check this out on your own and post your findings here in this thread.
Basically just filter for all hands in your database, this will tell you exactly how many hands you have played. Once you get this number, multiply it by 5% to get your expectation (again, simple math, no variance stuff, etc.). Then, go to PT4, filter for the following hole cards (AA-TT, AQs+ and AQo+ which is exactly 5%). Compare the two and post the computations in this thread.
As I expanded on this experiment, I did the same thing for top 10% of hands and then top 20%. My 10% and 20% numbers were within 1% of what the numbers "should" be. It is just at the top where things are weird.
My winrate, when I am dealt a top 5% hand is 340bb/100+ or 3.39bb/hand. As you can see, this period of variance has a massive impact on things.
So, those of you who go through the process I have described above and post your findings, I thank you in advance and I look forward to seeing the results.
I promise, this is not a bad beat post or "the site is rigged!!!!!!" post. I understand we will run below expectation. I am just curious to see where we all fall in the spectrum of things.
Thanks again
Loading 17 Comments...
I dont have big database(dont see any reason to keep old hands) but after 270k hands:
I got dealt AA 7% less often than I should(0,42% instead of 0,45%)
I got dealt TT+,AQ+ less often than I should(4,72% instead of 4,98%)
I got dealt 88+,AKo,KQs,AJs+ slightly more than I should(5,281% instead of 5,279%)
Now I now why I have long bad streak. Pokerstars is rigged!
lol RIGGGGGGED
Thanks for your input
The probability of being dealt (TT+, AQ+) is 62/1326 = 4.67%, so you actually got dealt a premium hand slightly more often than you should have.
AA (245), kk (247), QQ (236), JJ (277), TT (253), AQo (481), AQs (155), AKo (470), AKs (174)
Total: 2538 (4.4%) out of the top 5%.
57,113 hands.
You should consider standard deviations.
I'm not a math guy so I can't help you with that, unfortunately.
Many moons I've reached out a statistician asking for help to do custom stuff. Anyway, I've showed my PT4 bell curve graph where I was "extremely unlucky" with my sets and he told me that, in statistics, an unlucky event is something that occurs at least -4 std dev.
Probably you were unlucky but not that unlucky as you might think
I would recommend playing around with this tool:
https://www.primedope.com/poker-variance-calculator/
Kind of eyeopening (depressing lol) what actually can happen in poker even over decent samplesizes.
Oh yeah, spent some time here for sure LOL
How do variance calculators measure softness of a pool? Maybe someone is winning at 3bb because of their own leaks. I think in a soft pool it is hard to lose over 100k hands if you are playing your A-Game. Think a lot of this variance just comes from tough pools. Phil Hellmuth for example claims to win 24 out of 25 sessions but put him in a line up of pros and he can't do anything right. I don't fully trust variance calculators because it doesn't account for so many bad players in the pool.
RunItTw1ce
I guess the softness of the pool could be taken into account by changing the standard deviation. So if we have more of a passive pool we could drop it down (maybe like 70) and if we face a pool that splashes around all the time, we could set it to 110 more....
We could also change our winrate as we expect to win X more in a soft pool. Both changes of settings will have an impact on the variance.
I agree with standard deviation being a big part. Changing win rate sounds too manipulative in order to try and control variance.
I believe this topic falls into the cat. of variance. there is a clip here on RIO where they are discussing variance, and to have a statistic with 99% accuracy you need something like 6.5mil hands. You might get the correct % of top 5 if you double the sample size.
I also notice a "trend when i play" which i believe is variance as well. Im running either ultra hot, or ultra cold.
Ultra hot or ultra cold--I think this all the timel. Either the pool always has it and the worst possible runouts happen all day or everyone is an overcalling fish and I'm hitting gutters and flopping flushes. I know that's not true but it sure feels like it sometimes
The range you described, (TT+, AQ+), is actually only 4.67% (62/1326 combos).
862844 x 4.67% = 40344, and you got 40199. You got a premium hand 4.659% instead of 4.675%, which is well within normal variance.
You can use a binomial calculator to see how lucky/unlucky it is. There's about a 23% probability that you'd be dealt less than 42000 premium hands out of 862844 hands.
I agree with you 100%. I simply typed "5" into the box in PT4 and that range is what it hit on. Need to be more precise in the calculation since the fractions are substantial as the sample grows to larger numbers. Thank you sir
It's quite bad to focus on these things and blame circumstance as it takes away agency. Variance at small stakes Holdem cash games is very manageable if you have the proper soft skills and poker skills, so working towards that should be the focus and not looking for reasons or a story about what happened.
I do this sometimes if I’m losing and then make a correction for it in my winrate. Variance is really pretty wild, so I wouldn’t worry too much about it. Basically the next set of hands is independent of the previous set so it’s not like it has any impact on the future.
Nice to see you in the forums.
Be the first to add a comment