Ok, like I mentioned in an earlier post, I have a 40,000 hand database for my stars tourneys dating back about 1 year. I pulled out all of the hands that I was All-in and called - in other words, playing for my tourney life.

I've always had the theory that there is something significantly different about Stars from other sites. I win on all other sites except Stars. I've taken lots of abuse on these boards for my position that when Stars gets you down, they want you out.

Here are the the results of my analysis. I swear that I have taken every effort possible to be as accurate and fair as I know how.

All hands where I was allin and called preflop:
Hands: 223
Calculated Equity: 46.30%
Achieved Equity: 37.22%

Coin flip hands where I was allin and called preflop (43%-57% equity)
Hands: 56
Calculated Equity: 50.16%
Achieved Equity: 33.93%
Standard deviation = 6.317

40/60 hands where I was allin and called preflop
Hands: 17
Calculated Equity: 50.04%
Achieved Equity: 47.06%

Clear favorite hands where I was allin and called preflop (>60% equity)
Hands: 55
Calculated Equity: 74.06%
Achieved Equity: 70.91%

Clear underdog hands where I was allin and called preflop (<40% equity)
Hands: 95
Calculated Equity: 27.27%
Achieved Equity: 17.89%

Obviously, I don't have enough information to say anything other than "I run bad on Stars".

A few other things that I can say:

Coin flips: Jeez, no wonder I suck on Stars.

40/60 hands: I run slightly worse than average.

Clear favorites: Guys catch up a little more than they should.

Big Underdogs: I catch up significantly less than I should

It's actually kind of nice to see these numbers. I can certainly rest a lot easier knowing that I am "under the curve" when it comes to races. That's a much better answer than "I just suck at poker." I think it would be impossible to overcome these startling statistics for any player.

Now, why doesn't someone (MINDWISE) come up with a "how you running" statistic on some of these poker databases and hand history analyzers. I would really love to see about about 100,000 allin hands from 100 different players analyzed and see how the short stack fares.

THEN, I could finally have some peace about Stars' RNG.

PS: Please leave the "Sample Size" comments out of the discussion. Obviously, I'm aware that I can't PROVE anything. I was only trying to analyze MY play and identify areas that I could improve MY results. I gotta figure out how to compute my results when I call the allin with a bigger stack. I suspect that my numbers will then be skewed toward the positive. Tinfoil hat anyone???

Okay I edited my initial response. Basicallly, cliffnotes is:

1. check re-check your data anlysis, my guess if you did something wrong.

2. if you re-check it, and get the same numbers, have someone else re-check it besides yourself (in case you are doing something wrong and not realizing it) preferably someone who has taken college level courses in statistics and probability etc, and is used to crunching this out.

3. if you do all this and it still comes up with the 6.3 standard deviations from the mean figure, it DOES NOT mean pokerstars is rigged. There is somewhere around 1 chance in half a billion for this to happen over that 50something hand sample. However pokerstars has like 5 billion hands a year so there are about 100 million samples of 50-hand samples on pokerstars. This means that there is roughly a 20% chance for something like this to happen to someone per year of pokerstars existance. So roughly once every five years it is mathematically expected for something like this to happen according to statistics and probability.

4. Lol, you run fucking horrible. Lol.

<span><span>RE-</span>EDIT*** Apparentlyawwnutz's std dev # was not how many std dev's he was from the mean,but just the std dev. This makes my entire analysis worthless.</span>

-bfactor

Yeah. I've thought about calculating all-in stats like you have done here...but then I always think..."damn, it's not going to improve my game by finding out if I run bad or not"

JUST KEEP PLAYING...phuck these kind of stats

yee, phuk them.... but is really interesting, Im sorry I didnt collect my HH s in the past year, but a bit makes me think when all my HHs last hand I have some 75 /25 or 60/40 all in what I should win most of the times, and I dont...

40,000 is a pretty decent sample size, but in your subcategories are these?

Coin flip hands where I was allin and called preflop (43%-57% equity)
<span>Hands: <span> 56</span></span>
Calculated Equity: 50.16%
Achieved Equity: 33.93%
Standard deviation = 6.317

I realize those are 56 random hands out of 40,000 so statistically you should be able to draw significant conclusions from them, but when we are talking about poker and variance, in a sample that small one bad run could throw your numbers way off.

I'm no math whiz, but it seems like you HAVE run bad on stars over this sample, but is certainly no grounds for impeaching the RNG.

Ok here it goes for the coinflip one.

I am going to use a binomial probabilty distribution function using his numbers, as its poker you can either win the hand or lose the hand. It is binomial. It isn't normally distributed. Judging by the 56 trials, and .3393 equit, I know he had 19 wins, 56 losses, (19/56=.3393) when his probability of winning was .5016.

using handy dandy

1. binompdf(numtrials,p,x) - binomomial probability distribution function, <blockquote> numtrials = number of trials,
   p = probability of success, and
   x is the number of successes.
   It is equivalent to P(X=x), where P is the binomial probability distribution function P(x)=C(N,x)p^x(1-p)^N-x</blockquote>

bust out the ti-83+, binompdf(56,.5004,19) will be the probability he ran like he did.= .00509 or half a percent. not fuckin 6 standard deviations 5million to one. Only approx 200to1. If he included a bigger sample size he would undoubtedly get much closer. You people are all clueless I was going to sleep but seeing a moron like awwwnutz made me brush up on statistics and I feel so much better for myself even though I dropped out of college.

I just want to clarify that the binomial probability distribution function I used just told us exactly the answer to awwnutz question. "What are the odds that I only win 19/56 hands with my equity at .502" And the answer is .5% of the time he will win only 19 of the 56. If he ran exactly on average (28/56) that would only happen approximately 10% of the time. If he were going to find out if stars is rigged he would have to do uncertaintys and whatnot and I'm way too lazy for that. I'm just pointing out the retardation of people thinking that this 56 hands from a 40,000 sample means anything. Its perfectly standard/normal just think about it, everyone has gone on stretches where they seem to lose every coinflip.

You people feeding into this fools nonsense "lol 6.3 standard deivations 5mill to 1 ROFL u run bad, are just feeding into this nonsense"

Awwwnutz is bad at poker, and bad at statistics.

I'm going to bust out the ti-83 for a better answer to the question. ZOMG STATISTICS ARE COMING BACK TO ME

THE QUESTION WE ARE LOOKING FOR IS WHAT ARE THE ODDS THAT he wins AT LEAST 19 of the 53 showdowns.....SOOOOO SOOO SOOO WAIT FOR IT I NEED TO USE THE BINOMIAL CUMULATIVE DISTIRIBUTIVE FUNCTION. So in order to do that I'll get the odds he wins from 0-19 of the 53 and subtract it from one. That is the odds of awwnutz running as he did or better.

1-binomcdf(53,.504,19) and that will tell you how unlucky mr awwnutz is.

can't believe i'm remembering this shit now.

<span>THIS POST EXPLAINS IT ALL</span>

I'm really bored.
This is my last post and I will explain it all fully and use your total hand calculation and show you a better calculation.

The 40,000 hand sample means nothing. you didn't get an SRS, poker hands aren't normally distributed. The proper way to determine this is using a binomial probability function. There were 223 trials and your equity in those trials was .3722. .3722*223=83. Therefore we have the binomial function set up. 83 successes, 140 failiures, .4630 probability of success per trial.

What are the chances he runs at this exact rate over this sample?
binompdf(223, .4630,83) = .0012945. 1.3% of the time he will have this exact trial over 223 hands! hardly 5million to 1 you morons!

What are the chances he runs WORSE than this clip over this sample.
binomcdf(223,.4630,83) will tell you the probability he wins less than the 83 times he did. = .0038084176.

3.8% of the time he will run worse than he did here! I mean its obv a bad run of cards but not even near the worst run he'll ever see if he's played many hands!

Of course this means that he'll run better during this sample 96.2% of the time. So its a bad beat, bad run of cards, but nothing disgusting/bad about the RNG. We see hands that are less than a 3.8% of favorite win all the time everyday in poker.

This whole notion/idiocy brought up by awwnutz pissed me off tilted me and now i'm gonna have no sleep for the warmup fuck it.

This was my own thing I did manually a few months ago on FullTiltPoker during some $60 Turbo SnG's. Here's the image and link, and I'll summarize below. The sample size was 900 SnG's played over a span of two weeks, and there were over 2,500 all-ins. Long story short -- there was nothing out of the ordinary.

http://s36.photobucket.com/albums/e4...ngtotals-1.jpg



Summary of ALL-Ins

80/20+ -- WON: 229 LOST: 33 = 87.4%

70/30 -- WON: 300 LOST: 139 = 68.3%

60/40 -- WON: 269 LOST: 176 = 60.4%

RACES -- WON: 195 LOST: 183 = 51.6%

40/60 -- WON: 158 LOST: 232 = 40.5%

30/70 -- WON: 104 LOST: 304 = 25.5%

20/80 or less -- WON: 31 LOST: 173 = 15.2%

Overall

TOTAL ALL-INS: 2,526
Money in ahead: 1146 = 45.37%
Races: 378 = 14.96%
Money in behind: 1002 = 39.67%

NOTES

Even this sample size is extremely small, not to mention the way I'm calculating the percentages here is not very accurate. Basically, I'm rounding a 66/34 situation up to "70/30", etc. It's highly probable that there may have been more 27/73's that got rounded to 30/70 than 73/27's that got rounded down to 70/30, etc. The important thing to note here, is that the overall numbers are reasonable. It looks like I ran very slightly -EV over this sample size.

My base ROI% over these 900 SnG's was +4.13%. I'm guessing my true expectation over this sample size was ~+5.0% ROI.

David

I feel 10% dumber after trying to make sense of YTIU's calculations. Smart guys.

YTIU said it right in his own way...

just a simple observation YTIU. when u have a large enough sample we almost always assume that it can be approximately distributed normal - and if it comes from binomial its mean will be np, and its s.d. will be np(1-p)

just a simple comment..

the Expected Equity is about 50%, and the actuall is 33% with the s.d. being 6%. Empirically, u expect 95% of observations to be within +-2 s.d. (38-62), and 99% of observations to be within 3+- s.d.'s of the mean (32-68) .

So yes... for your limited sample size this is way on the unlucky side but it is quite possible. For every 100 players with similar sample sizes in conflips between 1 and 5 run similarly to Awwwnutzzz.

and yes, i also believe Awwwnutzz is right when he says that its difficult for anyone to overcome these stats... its simply almost impossible...

didnt read responses. you might run awful on stars but it has nothing to do with the RNG. get over it.

Bfactor,

I, like you, took two statistics classes in college - A's in both.

The standard deviation is 6.3 in the coin flip analysis, I'm not saying my results are 6.3 Std dev's from the mean. I am actually 2.56 standard deviations from the mean which equates to me being only in the bottom 1.5% of players who run bad.

Thanks for the feedback.

EDIT*** SORRY MY NUMBER WASN'T BETTER DEFINED

Lenny,

My only point about my sample size is that this is all I have for the results that I have. It took 40,000 hands to get this many. Unfortunately, I run so bad on stars that I am allin about 1.3 times per tourney before I bust. I'm not sure that one person can ever get the hands necessary to have meaningful data. That's why I encourage someone with access to more hands to run some numbers.

I'd like to say you are what is wrong with the board because you are an ignorant old fool who immediately tries to search for other reaons for his lack of success instead of immediately looking towards himself.