You are a beast Noah.

Stars and FTP, since you are definitely going to read this thread, why didn't you ever do this?

BTW, the cheating occurred on FTP both before and after the original investigation. In other words, the original investigation on FTP should've found it but didn't AND after they were investigated, they continued to cheat:

40putts's 3-bet % vs. Kinetica in CAP games before 03/01/09: 1.4%
40putts's 3-bet % vs. Kinetica in CAP games after 03/01/09: 1.8%
Kinetica's 3-bet % vs. 40putts in CAP games before 03/01/09: 1.1%
Kinetica's 3-bet % vs. 40putts in CAP games before 03/01/09: 0.9%

On Stars the picture is a little less clear:

knockstiff's 3-bet % vs. LittleZen before 03/01/09: 5.2%
knockstiff's 3-bet % vs. LittleZen after 03/01/09: 3.2%
LittleZen's 3-bet % vs. knockstiff before 03/01/09: 2.9%
LittleZen's 3-bet % vs. knockstiff after 03/01/09: 1.3%

I dunno why knockstiff's value is so much higher before 3/09. PTR doesn't have many hands on him then (~67k total, the statistic only has 502 hands and the 1SD error is ~1%), it might just be random error due to the fact that PTR hadn't been mining stars for too long. It's still a decent amount lower than his average 3-bet % of 7.92% over that time period.

tozzy,
It's a long-ass post, so I know people are gonna miss ****.

Does this answer your question?

Also, LittleZen was shortstacked for about 2/3 of the hands he played. When he's short stacked, he should in theory just view everyone as an open raising and a calling a shove range and nothing else. Doesn't matter what stack size they have except to the extent that it changes those variables. That argument doesn't work for knockstiff, but the one I quoted does, plus the fact that LittleZen was softplaying obviously makes everything look way more suspicious, plus the fact that the same stuff was going on in FTP CAP games (where stack sizes aren't an issue).

i have recently talked to a fellow player who is playing in a lot of draw games at pokerstars who plays a lot with other people at the table from the same town. he also plays a ton and claims to be a fairly big winner (i doubt that "big" from playing him though). when i asked him if he was softplaying or colluding with the other guys he knew, he answered, "no, we just dont bluff each other" and was very confident he wasnt doing anything against the rules.

lol.

they allow softplay which is very likely from the data. i dont think they have only idiots sitting there analysing data.

It'd be cool if a stars rep could 100% confirm, but I'm like 90% sure that Stars does not allow softplaying except in heads up pots (where their policy is wrong).

Noah,

please bear with me as I dont yet fully understand.

Is it possible that in those 30k hands you used to analize LittleZen's behaviour when playing knockstiff, both of them were shortstacking more than, say, 90% of the time?

If so, do you think this could explain some of your results?

Second remark (dunno if youve read my edit)
When plotting LittleZens tendencies versus oponents based on some individual stats, were you taking those individual stats from hands those oponents played against LittleZen or were those stats obtained from another source, maybe one with more data to improve accuracy of those stats?
If this is actually how you made those graphs, wouldnt that be a methodical error since you must assume LittleZen may very well have totally different stats on those oponents to base his actions on?

---

Actually forget the second one, I think youre right in assuming that any sample large enough for someone to base his actions on should not be so much different from that oponents real stats (or, rather over a v. large sample).

Well done, Noah, very very well done.

I sincerely hope Stars take notes of how your investigation went and start writing scripts to check for such discrepancies.

Another good question.

The stats are from PTR's full database. This might increase the error bars a bit on the very low sample size stuff because of the potential that stox might have been seeing different stats there, but it's not too significant.

The worst case scenario is one guy who only had just over 2k hands with stox in PTR's database. 2k hands is obviously a really small sample for stuff like determining winrates, but that's plenty to get a good idea of PFR/VPIP and usually a decentish idea of F3.

Plus, since this is all based on 49 opponents, I can afford little screw ups like that.

No. LittleZen shortstacked (less than 25 BBs) something like 2/3 of the time in PTR's full DB and almost exactly the same amount of the time when he played with knockstiff.

Off the top of my head, I think knockstiff shortstacked like 40-50% of the time and was shortstacked about the same % of the time in hands with LittleZen.

Addressed above. You're right that I screwed this up, but it won't change anything.

I'll be the umpteenth person to say how impressed and grateful I am, Noah.

Also, this phrase gets used to much, but this is truly sickening.

Incredible stuff Noah. Very impressive.

Cool. Thanks for the explanations.


There is, however, a really small chance that the accused people here are in fact innocent, so we should think about ways to actually prove it.
Now, what about if stars or ftp could review all hands played by those players, looking for patterns regarding their actual hand (I assume they have HHs with all players' hands visible, right?).
Say when LittleZen is shortstacking and villain with more than 75 BBs (or make it 100 if you want) is openraising the BTN first in to <5BB, he shoves AQs in the BB >90% (I think it's gonna be even more than that, but it doesnt really matter). Now when he is facing knockstiff, he is shoving a certain %<90.

Question:
How much difference would there have to be and over what sample for collusion to be proven?

owned so hard