(Sorry for the slow response. I've been out of town.)

Thanks for these links. I've been to your site a few times but hadn't gotten to these pages. For anyone following along, Spadebidder ran an analysis on 150 million hands from two sites to test Barry Greenstein's theory that card removal effects could sometimes cause AKs to become favorites over most pairs, depending on how many people fold prior to the AKs hand having to act.

I'm posting the graph of the results here:



This shows that in the huge majority of cases, the change in equity produced by card removal effects, at least in the case of AKs vs. 44-99, is around 1% or less. I would argue that a change this small would not be enough to noticeably affect an equity analysis for any one person. Random noise would, I think, overwhelm these small card removal effects in a sample size that any one person is capable of producing. The graph shows this in the difference between Site A and Site B, especially in the last two data points, although I'm sure we're running into sample size issues there.

I would also argue that AKs vs. pocket pairs represents one of the biggest equity shifts you'll get from card removal effects. Maybe the biggest, which is probably why Barry used it as an example in his book. For most other situations, I would expect card removal effects to produce even less of an equity shift.

Even assuming that card removal effects are enough to invalidate "luck" calculations, since we're aware of them shouldn't it be possible to incorporate them into our luck calculations to correct for the effects? I would think even a street by street equity analysis could be accurate if such a correction were applied properly.

I agree that for preflop all-ins the effect is not significant for any single player (and barely matters in the aggregate). That's why luck analysis is usually confined to preflop all-ins. My point was that post-flop the effects are compounded. If more than 2 players see the flop and then some of them drop out before the end of the hand, we've now seen more cards for players to base their decisions on. That's why post-flop all-in analysis is more biased. There are some other good explanations on this forum of why street by street equity calculations are not accurate when done with incomplete information (not knowing what hands folded post-flop). It's a pretty safe assumption that almost always those folds are decided primarily based on the relationship of the board cards to the player's hole cards. Meaning they are anything but random. Yes, maybe they tend to cancel out over time to a somewhat random distribution, but I doubt it. We could easily think of play style examples where a particular player would have highly biased post-flop results. But the main point is that once you allow decisions into the mix, we are no longer measuring luck.

By the way - you hotlinked the image from my server. Should be ok.

A little late, but I appreciate the effort.

Can't they hire better shills?

The shills are silent

Since anyone who does not blindly espouse the rigtard's cause is immediately labeled a shill I very much doubt it will stay that way for long.

*sigh... but when a good player gets allin AAvs22 and wins you are then running above ev. pro allin $100-AA vs noob allin $100-22. The pro won $100 but was only intitled to $82.21.

So no, a pro won't allways run bad.

Yeah, sometimes my jokes suffer from the "humor removal effect"

But since we don't have complete information, is any of this more than speculation and educated guessing? Wouldn't it be better if we had the missing information and could measure the effects directly?

We know no poker site would give out hand histories with all hole cards exposed, but that doesn't mean it's impossible to get them -- they'd just have to be created. There is an open source Texas holdem simulator that works on Windows, Mac, and Linux, and it comes complete with server for a poker site and client software for players. Play can be against other players or against pre-programmed opponents. Here's a short youtube video showing the somewhat ugly interface:
PokerTH in action

It's written in C++ and includes a number of AIs you can play against. The AIs are probably for people who don't know much about the game and want to practice against the computer before playing against other people. It doesn't really matter why they're there, though. The important thing is that it's all there and it's open source. That means this whole system is basically a framework for any kind of Texas Holdem simulation anyone would want.

If a million hands are needed with complete information from 9 players who never change their styles, it's easily done. If we used 9 identical players who never tilted and never changed, that would tell us, for example, how much of a factor variance truly is in this game, and how big of a sample size we really need before variance is no longer a factor. Datamined databases, with hand histories from human players, have too many variables (human emotions, human learning, human distractions during play) and not enough information (we can't see all the cards) for certain kinds of studies.

Of course, this would be useless for testing the "riggedness" of online poker, but I think that's only a tiny part of your study, anyway, and probably the least interesting. A simulation would allow you to control variables that can't be controlled in mined data and would guarantee a sufficient amount of data no matter how much was needed for a given study.

I'm not trying to talk you into another massive project. Really, I'm not. I just wanted you to be aware of a resource that's out there if you'd like to look at it. I think you could do things with it that aren't really feasible with the mined database you're working with now. Plus, 95% of the coding is already done. Something to think about for the future.

Oops. A thoughtless oversight on my part. I wanted the image to appear in my post, but I was thinking about the text of my post while I was trying to figure out how to post an image, and bandwidth issues just never occurred to me. Hopefully this page will fill up and roll over before too much is chewed up. Sorry about that.

Heh. This one didn't.

Thought?

Full Tilt and Poker Stars can send you a TON of links to "prove" to you they are fair and have random shuffle programs etc. Even though the programs "I'm Told" do not burn a card, but lets say they are all fair and honest for the sake of this argument...

OK.

What are the real chance, that the sites themselves are Fair, but others have "hacked" into the sites and can see your cards, or they know the program the sites use and can predict the next turn and river card?

Wallace loses.

There is almost zero chance that anyone has hacked into any of the main sites and can see your cards. The chance that someone has hacked into your computer and can see your cards is much, much greater.

And there is zero chance anyone can predict the next card to be dealt (unless it's all rigged, of course).

What does burning a card have to do with anything? I think you are confused about a few things.

in before burn card theory goes to ****

I sometimes feel that it is "rigged" or that a superuser is playing the game due to some of the horrific and unexplainable actions that are done on occassion... usually Sunday mornings/afternoons. Played today and shoved with KK with a decent stack and was called by 45o (preflop) and they hit a straight... things like that in Cash and SNGs will start to make you think that there are superusers playing on the site... ho hum... just another day...

Burning a card after the Flop, and Turn is the way Live Poker is played.

I made the assumption most reading this forum had played Live Poker and not just on-line.

The point being, burning a card changes the outcome of the hand, and on-line sites such as Full Tilt and Poker Stars do not burn a card is all.

What does anyone else think about it OR did anyone else even know this?

You're looking at this the wrong way. The object is to get a random deal. Burning a card is done live to avoid people gaining an unfair advantage by spotting the next card to be dealt. This problem obviously does not apply to online poker. While doing a virtual burn would affect what cards are actually dealt for a site like pokerstars, it is irrelevant except for emotional reasons. For a site like FT which uses a continuous shuffle, then of course burning a card would be even more ridiculous.

As long as the cards are random, you should have no concerns. Burning is a red herring in the rigged debate.

Makes sense, thanks

From what I´ve read in this thread it seems very clear that no data analysis is good without full information (knowing all players hole cards). So you cant really know if the deal is rigged without knowing all hole cards

The rooms wouldnt give us the hole cards because of confidentiality matters so you cant go any further in the data analisys.

Please let us know which posts make this clear.

Look, obviously it would be great to have all the hole cards. No one would argue that that is the most complete analysis. But as spadebidder has explained several times now, there is still quite a lot that can be deduced without having all the cards.

From what Ive read in the spadebidder site it is very simple to make analysis of PF AI situations but turn and river action is very difficult to put in the randomness limits because theres a lot o card removal effects and personal decisions taking place.

You seem to believe that not burning a card would make a game unfair. This is incorrect. Burning cards is only done as an anti-cheating measure so that the dealer can't as easily cheat and so that people who mark cards still can't see what the next card is since it is underneath one that will be burned.

Obviously at online poker there is no need to prevent either of those things so burning a card is unnecessary.


As fr the rest of what you said .... I think it's very unlikely.

Rubbish.

You could very easily show a rigged deal without knowing any hole cards.

If that isn't immediately obvious to you not only do you not know anything about probability maths and stats you would appear to know very little about poker.