in early december i ran white hot at 5/10 and I recall thinking that it was likely one of the hottest stretches of my career. I ran at 28ptbb/100 after 5k hands of heads up.

The hottest I EVER ran was a stretch at 2/4nl 6max almost a year ago where I ran at 21ptbb/100 for ~45k hands.

here's the sick thing about all this:

only 30k hands for this guy is $3million at that winrate and those stakes.

josem: just for comparison, how improbable does your model say this 28PTBB/100 over 5k stretch would be? i feel like i've run about that good over stretches about that long more than once, so if your model says its close to impossible theres a problem.

Nothing, I mean nothing is "obvious" at all about this case. 3000 hands is a tiny sample especially for someone playing ridiculously loose. His variance will be far higher over 3000 hand samples when compared to much tighter players. It also seems like youre assuming his play being "unprofittable" as proof of something. No one is saying this guy is even a winning poker player over the long term. I think it's pretty clear he was either a fish on a sick run or something shady is going on.

56bb/100 would is just over three standard deviations above the mean.

According to my sample data, one would expect around 99.7% of such hand samples to be between -56bb/100 and +56bb/100.

That is, on average, you're looking at something in the vicinity* of such a run once every 333 samples - ie, on average, once every 1.5million hands. Instinctively, this feels about right to me.

Keep in mind, though, that a poster on here is more likely to have hit this point because they are more likely to be a winning player plays better than average (by definition), and thus would have a 'mean' that is higher.


I would imagine that a number of 2p2'ers have exceeded 56bb/100 over 5k hands a number of times. Depends how much each has played, TBH.


*do the full math if it precise accuracy is important to you

I thought that the fact that the suspicious player was running more or less at expectation when all-in suggested that the variance issues here are pretty small.

Superusers should play limit, then no one would care. A9 high valuebet? Standard!

Here's another pretty graph I've just made that shows the distribution of winrates in bb/100.

It seems to indicate that if winrates are not distributed in a traditional, bell-curve, normal distribution, that it is sufficiently close to draw similar conclusions. Even if I was wrong on this point, when something is as "out of whack" as the suspicious player, it is worthy of further investigation.

UB should look into this, because they have the data to determine this conclusively.


-BB/100 has been rounded to the nearest whole number for the column charts
-the yellow bar on the far right corner is the suspicious player at UB
-the sample is 330 players, who have all played between 2,500 and 3,500 hands.
-any questions, just ask

In the graph that I've just posted, 56bb/100 divergence from the mean is shown to have occurred twice in 870 samples - pretty much on target.

Josem, I think that for future graphs you should use the same amount of hands for all of the players in the sample, so in this case either 3,000 hands or 8,000 hands, we know the winrates for both samples.

correct me if i'm wrong, but over a sample this size isn't having higher variance more helpful to your chances of booking a huge win than having a higher winrate?

OK - done. The graph is now players who have played between 2,500 and 3,500 hands (inclusive).

That sounds likely to me.

From the Sklansky bucks graph posted, however, I understand that the suspicious player was running at expectation on the stuff that has been analysed.

I don't understand why people keep saying that he is running hot when there is no evidence of this.

I didn't look at all the HHs (all 3000 or whatever), but all-in luck shouldn't be the only thing considered when it comes to his variance. If he's loose and bad he'll be putting a lot of chips in with hands that may not even go to showdown. Combine that with the fact good players viewing him as a fish are going to be trying to build pots vs him with non-premium hands. So his perceived looseness/fishiness is going to lead to a lot of big pots as well. If he's getting 2pair vs TP or TPTK vs TPGK a lot over those hands he'd still be getting the money in good (thus not lucky). He could just be running normal/good in all ins, but running very hot in "cooler" type hands. If those 3000 hands capture such a run it wouldn't be impossible for such a player to have a super high winrate. Is it really unlikely? sure, but outside the realm of possible?

Of course it is "possible" for someone to theoretically win at this rate and have a series of obscure and unlikely reasons to be winning.

However, it seems unlikely, and adds further weight to the need for UB to investigate and come clean about what happened here.

This is not something that is, on the evidence to us, like hitting a 1-outer. We're talking about stuff in the vicinity of hitting fifteen consecutive 1-outers.

Great news guy. http://www.mypokerintel.com has agreed to send us what they call a 'high-level summary' of the play of nionio. In the email they also stated that they are more than willing to send me the 8200 hands nionio played however there was no attachment in the email. I presume I must email them again to get the hand histories. I want to give a special thanks to http://www.mypokerintel.com for being so helpful in this matter. Anyways here is what they sent me, when i receive the final 8200 hands i will post immediately

Edit: link didnt work, tryign my best here

http://spreadsheets.google.com/ccc?k...j3f21JZ325s8dw

Edit: im an idiot

A few points to be made. First, our data is valid, and hopefully Taylor Caby would still vouch for that, as we went through an extensive exercise with him on this issue a while back, comparing some of his results to PT.

Regarding the above MyPokerIntel.com "Leaderboards" pasted all over this thread, those stats reflect play at all levels. Recognizing the inherent flaw in comingling .02/.04 results with $25/$50 results, we have broken these same ranks down across semi-arbitrary limit lines as "micro" "small" "mid" high" and "top", keeping the "overall" also. This is data we have meant to expose on the site, but the leaderboards are by far not the point of MPI and hey, we have day jobs. Thanks to some of the posts made by those who don't seem to know what MPI is about, we'll be bumping that task up on our "to-do" list.

In short, at limits that matter (MID and up), getting over 26+BB/100 doesn't seem to happen over the 400M hand sample that we have. Yet, with the player in question on this thread, the 60+BB/100 was achieved ALL AT TOP STAKES, over 8,000+ hands, not 3,000.

As for the crazy high BB/100 at the low limits, the hand histories indicate that players screw around and/or "transfer" (by losing HU) relatively large (for the stake in question) sums to each other, which of course is captured by our Intel-engine as huge, massive BB/100 "wins".

BTW, we have also agreed to help Trambopoline et. al. with looking further into this by providing whatever they need; and if its data and/or HH, we got it.

- http://www.mypokerintel.com

FYI - The top 10 BB/100 for each limit strata:

TOP
PlayerName Stakes Hands BB100 Rank LastPlayed
worldsgrtest TOP 6,478 15.02 1 1/3/2008 0:05
lumbardo TOP 16,467 12.41 2 1/1/2008 15:12
applsgirl TOP 8,305 9.79 3 1/3/2008 1:35
dickholdem TOP 9,854 9.79 4 12/31/2007 16:34
swipesville TOP 10,829 9.28 5 1/2/2008 20:30
Mara10 TOP 17,099 8.40 6 10/11/2007 21:09
joes2828 TOP 5,367 6.94 7 1/2/2008 20:02
Senior Jota TOP 15,107 6.70 8 1/2/2008 12:39
WeeMaN TOP 7,857 6.39 9 11/22/2007 14:32
elpanchito TOP 22,783 5.98 10 12/20/2007 10:33

HIGH
PlayerName Stakes Hands BB100 Rank LastPlayed
Boots_TEXAS High 8,612 22.68 1 12/10/2007 3:59
matadoren7 High 5,658 13.48 2 12/30/2007 17:12
chrisk276 High 12,838 12.58 3 12/2/2007 6:30
PiMaster High 9,537 12.14 4 12/6/2007 23:06
TheCatJuggler High 7,706 11.74 5 10/25/2007 8:16
WoodDizzle High 5,304 11.22 6 11/17/2007 4:25
geormiet High 5,990 11.13 7 12/23/2007 21:43
MUCKEMSAYUHH High 10,367 10.73 8 12/29/2007 22:50
EDGARIUS High 6,278 10.65 9 10/26/2007 14:08
Lord_Erik High 14,791 10.57 10 12/7/2007 16:06

MIDDLE
PlayerName Stakes Hands BB100 Rank LastPlayed
CXXRA Mid 5,209 26.61 1 11/25/2007 4:52
James_NYC Mid 5,134 19.22 2 1/3/2008 0:56
SF_Squeezerr Mid 17,848 17.20 3 12/17/2007 23:32
Buddha_Sack Mid 5,232 16.29 4 1/3/2008 6:46
Unagi80 Mid 20,453 14.46 5 10/20/2007 23:42
mckuin Mid 5,377 14.29 6 12/24/2007 20:12
haugsrud Mid 5,318 13.88 7 12/23/2007 17:10
zandourizingo Mid 17,441 13.35 8 1/3/2008 1:58
TomfooleryU Mid 6,131 12.72 9 1/1/2008 1:50
gimme_dat Mid 13,580 12.28 10 12/30/2007 15:26

LOW
PlayerName Stakes Hands BB100 Rank LastPlayed
Cheddah Low 9,395 274.17 1 1/2/2008 0:44
Bear Rabbitt Low 15,418 26.64 2 1/1/2008 0:30
badgammon Low 10,768 23.96 3 12/31/2007 21:09
AKA Chicago Low 14,770 22.67 4 11/23/2007 19:47
Impact3 Low 12,290 18.84 5 12/16/2007 5:40
Mr Gobbles Low 9,168 18.51 6 12/28/2007 18:45
kidpoker4848 Low 8,464 18.30 7 1/2/2008 4:58
LatestLines2 Low 7,776 18.18 8 11/19/2007 5:23
RussellAces Low 6,641 16.29 9 11/13/2007 15:13
jimmie23 Low 8,949 15.78 10 12/4/2007 4:08

MICRO
PlayerName Stakes Hands BB100 Rank LastPlayed
Johnny Dub Micro 18,142 234.34 1 11/10/2007 18:56
J24S Micro 6,167 64.22 2 12/5/2007 4:21
push bad hand Micro 14,545 62.07 3 1/2/2008 21:23
saltymctilt Micro 9,136 61.80 4 1/3/2008 16:33
Acey Ducey Micro 19,837 57.30 5 12/25/2007 4:47
Pocket J0kers Micro 8,496 49.51 6 1/3/2008 13:11
holddunutz Micro 7,679 49.03 7 11/20/2007 3:56
phobmeisterfl Micro 8,433 39.32 8 11/24/2007 8:24
mrpresident84 Micro 5,811 38.95 9 12/7/2007 12:05
pokemeup Micro 7,362 37.90 10 1/3/2008 3:06

OVERALL (the flawed method)
PlayerName Stakes Hands BB100 Rank LastPlayed
Legiao Overall 46,633 696.31 1 12/31/2007 3:00
Toesanders Overall 105,871 356.64 2 12/30/2007 6:45
therentmoney Overall 13,884 38.50 3 12/18/2007 1:59
Pocket J0kers Overall 11,309 36.46 4 1/3/2008 13:11
Johnny Dub Overall 116,712 35.39 5 11/10/2007 18:56
CAM997 Overall 35,127 33.98 6 12/31/2007 6:17
Acey Ducey Overall 31,891 32.36 7 12/25/2007 4:47
phobmeisterfl Overall 10,232 29.00 8 11/24/2007 8:24
1967barracuda Overall 36,090 28.84 9 12/29/2007 23:11
justin scott Overall 57,208 28.35 10 1/3/2008 3:47

We're sorry, xxx@gmail.com does not have permission to access this spreadsheet.

The column on the spreadsheet (which was a quick and dirty dump) indicates Big Bet, so the big blind would be half that, as you indicated.

Yea, sorry about the link. Im terrible with computers so I just forwarded it to omniheart. Hopefully he can repost what i was trying to post.

Damn. I was definitely leaning toward superuser, but still kind of on the fence on this one. I think that pretty much seals it.

What really scares me is that UB and AP software were developed completely independent of each other right? So how many other sites have this capability? All of them? And how many others on UB aren't being as greedy as this guy and flying under the radar?

I suck at cash. But I'm pretty sure if I set my mind to it, I could look like one of the best soul-readers on earth, take high stakes for a pretty nice clip, and never be detected. Just play your best normal, then check the super user every now and then when you have a tough decision. But not always.

When you say "> 26 pt/BB doesn't seem to happen" - what is your sample size minimum threshold?

My 2 cents.. Its pretty obvious its a cheater, we should expect that it is a former UB employee, or linked to one.

He must have been pissed when potripper got caught.

Something I found fishy; when talking to Phil Hellmuth on UB today, I was bombarding him with questions. I brought up something I remembered him saying about 4 months ago. When asked if he was still going to stay with UB after the whole AP scandal, I specifically remember him saying like 'yes because ub is still safe' He also said that if he thought UB was not safe he would leave the site as a pro. When i brought this up to him today, his view was sorta changed. I asked him if he would continue to be a pro for UB if Nionio was confirmed 100% to be a cheater and he said 'yes i would if it was just some rogue employee'. Now im not saying Phil knows anything particular, but I did find this comment kinda strange. Not once did i mention that I thought it was an employee at UB, but from this statement I kinda got the feeling that thats what he thought. I dont know, just my thoughts

Has anyone tried to talk to anyone at UB? Is that even possible? They know who NioNio is. Would be nice to hear something from them.