cash and tourneys aren't anywhere near comparable and doing so is just a waste of time. also DNEGs is pretty terrible at cash, iirc.

Third prize is you're fired.

Most cases get settled, and Stones is likely to be willing to settle. They don't want the publicity of it going to court, and they can blame a crooked employee.

This JFK is going to have a hard time finding employment in a casino or elsewhere. It is likely this was his idea and he recruited the the worst and sleaziest grinder he could fine.

Really wish POS Postle would flip on Stones

I mean if we accepted the hand-wavy statistics ITT then he would be guilty beyond a reasonable doubt as well. The problem is that neither the mean nor the variance of his winrate are known or even estimable, and there are numerous fundamental issues trying to argue them in a quantifiable way. There's kind of this aura of arrogance ITT that if you don't buy into that argument that you "don't understand statistics" but as someone who's both spent years as a poker professional and as an actuary, there are many critical problems with this approach.

For example, even if you wanted to try to compare winrates between Postle and whatever unverifiable, uncontrolled data you have for live winrates elsewhere, it's easy to argue that because Postle has a high VPIP that it could severely distort the mean and variance of his winrate against professionals playing a fundamentally different style even if you could somehow control for all the other external variables (which you can't).

Edit: It's also worth pointing out that the circumstantial argument that if Postle was truly a great player that he would move up in stakes that has been repeated a lot is actually an even stronger argument against the comparability of winrates to Postle and others. If all the great players move up to high stakes, then there's a selection bias against pros in the low stakes that their skills are lower than what is attainable by others.

I believe they're hoping to find hard evidence either in discovery or voluntarily through Stones.

Having a hard time believing you show a winrate in poker with that second paragraph, just complete nonsense.

This is the kind of self-unaware ignorance I'm referring to, by the way.

The control group should be Postle in the streams pre-7/18/2028

More generally, the reality is with any defense case that you haven't seen a lot of the evidence and really don't know what defense you are going to put on at the outset.

This was a placeholder statement- it was something the lawyer could say so that the media would report he was denying the charges. But it doesn't have anything to do with what they will or won't do in court.

If they legitimately had a third party do a review it would appear they have a malpractice claim and defense.

That's not right at all. Most convicted murderers go to prison on circumstantial evidence.

And the amount of evidence that is going to be available to the prosecution is staggering. This thread just scratches the surface. Prosecutors may very well recover Postle's phone and the phone of his co-conspirator. They will certainly have several other camera angles we don't see, including both surveillance footage and b-roll footage. They may also have footage of people going in and out of the peep room. They may have the computers that ran the hole card graphics software.

And they're almost certainly going to gather a ton of witnesses who are probably coming forward right now reporting this or that suspicious thing Postle and his co-conspirator did.

While this case isn't a total slam dunk (yes, a defense lawyer can muck it up somewhat with bogus statistical arguments and the like), if the prosecutors decide to go forward this is going to be an extremely strong case that, in the hands of a skilled prosecutor, will more likely than not result in a conviction.

Still can't control for the players at the table, his skill development as a player, or the games and stakes being played. The other issue is that the data is being grouped only after the results are already known. By that reasoning we can similarly conclude that Fedor Holz and Justin Bonomo must also have been cheating in their respective heater years if we just draw the lines wherever they fit to advance the narrative we want. This isn't even playing devil's advocate at this point, this is just fundamentally bad statistical practice.

This is really a path to nowhere. The closest reasonable approach I've read ITT is whosnext saying that you could do some kind of hand breakdown to measure Postle's expected winnings against some kind of range. That still requires some hand wavy estimations, but this is the closest thing to an actual statistical argument you can make in a way that makes objective sense. And it would still be very nuanced.

I agree with this in principle. I haven't seen enough of a real analysis of the winrate and play behavior to make a statistical argument. Especially now that we know that initial spreadsheet was off. Add in the fact that we're looking at a sample size of what ... 200 hours? That's not all that big in live poker.


But when we apply the Eye Test to what's going on, it certainly looks like either he's cheating, or his dick can read people's souls.


It should be possible to go through every hand of the live streams and build a database of his play and hand results (just like we do with PokerTracker or whatever the current software is), before and after implementing the CTO strategy. That would give us a real winrate and some real data. Problem is that that is a huge task and it's difficult to rely on the graphics in the stream.

It's more than that. Having discussed famous cases with non-lawyers many times over the years, there's just a tendency that non-lawyers have to think that cases cannot be subject to any nitpicking at all if they are going to win.

You can see this upthread with the people positing how Postle will explain his crotch staring and other suspicious actions. "Oh, he'll just say he was watching sports" or whatever.

Every major contested case against a guilty defendant has some weaknesses. Sometimes a case against a guilty person has major weaknesses (e.g., Mark Fuhrman's use of the n-word). Sometimes there are less significant weaknesses (e.g., a somewhat important prosecution witness has a criminal record, or isn't very good in the face of cross-examination).

It's actually really, really hard for criminal defendants to beat the rap when they are guilty. Indeed, if anything the system bends a bit too far in the other direction-- e.g., juries are too quick to believe prosecutors and police officers, many of whom are less than fully honest with respect to what they say in a courtroom. And guilty defendants who do beat the rap usually do so due to a combination of governmental incompetence/malevolence (again, see Fuhrman, Mark) and brilliant defense lawyering that involves successfully poking holes in the prosecution's case WITHOUT putting the client out there to lie and get exposed (see Cochran, Johnnie). It's very, very, very rare for a defendant to beat the rap by taking the stand and telling a bunch of lies, or by a defense team getting the jury to believe a bunch of lies. The rare times when it has happened (e.g., R. Kelly's first trial) usually involve a jury pool predisposed to a not guilty verdict anyway.

I have virtually no doubt that he cheated. But I think people are turning to statistics hoping that if someone says a magic number that suddenly Postle is ****ed when in reality I would be far more comfortable on Postle's defense in this regard. It really comes down to discovery and Stones's investigation here.

There's also the financial records of the co-conspirators. If JFK is suddenly getting $50k in extra un-accounted for income that's a pretty big flag. Or one of the tech guys.

Disagree in many ways. The point of a regression is not to control for EVERYTHING. That’s called overfitting your model. If you want a better instrumental variable than pre- and post- 7/18 the. You can use whether the phone is on the table or not.

Yep.

There's also tons of more speculative, wild-card type evidence out there. I have a lot of sort of 1 percent chances that I would pursue if I had the resources of a prosecutor on this thing, but here's a couple of examples:

1. Did Postle and/or his co-conspirator owe any poker players or shady lenders significant amounts of money? Many of the poker pros I know are often borrowing and lending money from and to each other. Needing to pay a debt quickly is one reason a person might move from playing to cheating.

2. Was Postle and/or his co-conspirator ever involved in any sort of cheating at other casinos? This might be fruitful because he moved around the country so much. He was in Mississippi for awhile I believe, and he was definitely in Nevada, and he comes to Sacramento, CA. Why did he move there?

The point is, a well-funded prosecutor will brainstorm and make a list of 100 of these things, and maybe 2 will pan out. But those 2 might yield dramatic evidence of Postle's guilt.

Overfitting a model is not controlling for as many variables as possible, it is when you make the model overreact to variance in the observations in a way that assigns signal to the conditions in the observation where there is no signal. Controlling as many variables as reasonably possible is actually one of the best ways to prevent overfitting. It just isn't realistic.

But the players in the game, the stakes of the game, and the actual game being played are all extremely meaningful features. That's just one of the many reasons the classic statistics approach doesn't really go anywhere.

Assuming they will be able to subpoena the unused show footage, and (more importantly) the overhead security footage, they will be able to determine precisely all of the relevant statistics.

Agree on definition of overfitting. Stakes and type of game would be controlled for. Why wouldn’t it. There will also be a lot of overlap on players. The classical statistical analysis, if the inputs are correct, will be compelling one way or another.

Statistics are just too easy to demagogue.

All the other evidence is what is likely to convict Postle.

If you knew anything about poker you would know that Mikes win rate is in fact impossible for any human being to achieve over 10k hands, Poker is in fact a very complex game and can even be solved mathematically in theory. The only way you can achieve a win rate like Mikes over this sample size of 10k hands is with more information then your opponents (AKA: seeing your opponents cards). Technically when you have this ability your not really playing poker any more are you? And this can be EASILY shown by his win rate. In theory you could put a perfect Game Theory Optimal super computer against the players at stones and honestly the computers win rate would look NOTHING like Mike postles, thats because the super computer cant see the damn cards.
Again, Poker sites all around the world use this tactic to catch cheaters and bots ALL THE TIME.

Yes, it is extremely unlikely that Postle's defense could hold up in discovery.

Ok so from this and previous posts, I gather that

1) It may be somewhat difficult to get to criminal trial, depending on who's investigating, but if they did, there's a good chance for a conviction
2) A civil lawsuit is easy to get to trial, easy to find him guilty, but hard to decide how much plaintiffs should get paid.

Would this be an accurate assessment?