General Poker Discussion Poker Forums

Since when is the improbable impossible?

David Blaine makes the impossible improbable.

in a full ring game, the odds of observing two players flop set-over-set assuming ppl would play all pocket pairs is around 4400-1.

That is actually the odds for heads up play, not full ring. Not trying to nitpick or get off topic

Back to the topic...

2000-1 happens, 1325-1 happens, I agree with all of that. But I will keep shoving all in preflop with AKo and every time I get called by an underpair, I will document it. This is the key statement... If I ever run 5 SD's from the norm, then that proves that something fishy is going on. Do you at least agree with that?

That is actually the odds for heads up play, not full ring. Not trying to nitpick or get off topic

Back to the topic...

2000-1 happens, 1325-1 happens, I agree with all of that. But I will keep shoving all in preflop with AKo and every time I get called by an underpair, I will document it. This is the key statement... If I ever run 5 SD's from the norm, then that proves that something fishy is going on. Do you at least agree with that?

no its the odds that two players will see the flop and hit set-over-set in a 10-handed game. i didn't do the calculation myself but it is supposed to represent the times you would be standing and just watching the table...every 4400 hands two players will flop set over set.

anyway, regardless of the calculation, i would not agree that if you run 5 SD from the norm that something fishy is going on. 5 SD isn't impossible, its just highly improbable. there are lots of players, and trillions of hands have been dealt. its going to happen eventually.

and again, looking at just your AK hands is disingenuous. unless you can prove that there is a systematic bias for certain cards to fall or certain hands to win over an extremely large sample across many different situations, you have no evidence of a problem.

by your logic, it might be reasonable to conclude basketball players above 7'6" are cyborgs from the future b/c they are 10 standard deviations from the average male in height.

no its the odds that two players will see the flop and hit set-over-set in a 10-handed game. i didn't do the calculation myself but it is supposed to represent the times you would be standing and just watching the table...every 4400 hands two players will flop set over set.

So this is an empirical observation? Not the actual mathematical calculation?

So this is an empirical observation? Not the actual mathematical calculation?

i got it from a 2p2 thread where he makes a bunch of assumptions. the number isn't really important and its not something i'd ever use for strategy implications, so don't get too focused on it. here's the post if you're interested:

http://forumserver.twoplustwo.com/25/probability/probability-set-over-set-380320/

and again, looking at just your AK hands is disingenuous. unless you can prove that there is a systematic bias for certain cards to fall or certain hands to win over an extremely large sample across many different situations, you have no evidence of a problem.

If I took a dataset and said "Let me find some anamolies." That would be incorrect. What I am doing is saying "Let me make a hypothesis and perform an experiment", which is totally different.

Let me ask you this. Let's say, hypothetically, that you lived in an alternate universe where something were amidst with online poker. What would it take to prove it? Or, in this hypothetical universe, would they be able to fool you forever?

If I took a dataset and said "Let me find some anamolies." That would be incorrect. What I am doing is saying "Let me make a hypothesis and perform an experiment", which is totally different.

Let me ask you this. Let's say, hypothetically, that you lived in an alternate universe where something were amidst with online poker. What would it take to prove it? Or, in this hypothetical universe, would they be able to fool you forever?

it would take millions of hands (or whatever the statisticians deem is significant) where they found a systematic anomaly with a high enough p-value. needless to say that if trillions of hands of poker have been dealt and no such anomaly has been found yet, despite us having all the required tools, i would be incredibly surprised if you were the person to find it. thats not personal, i just don't see how they could have fooled us for so long when you can go on PTR and buy millions of HH's at a time these days, plus the sites have one of the same independent auditors telling us everything is legit. if you actually read how the RNG works and believe they are telling the truth i don't see how you could ever claim that it doesn't work correctly, and if hte RNG hardware works correctly, then the only explanation is that FTP/Stars/auditors are all colluding together to cheat players out of money, without a clear motive to do so.

its much, much more likely that you just ran bad.

i got it from a 2p2 thread where he makes a bunch of assumptions. the number isn't really important and its not something i'd ever use for strategy implications, so don't get too focused on it. here's the post if you're interested:

http://forumserver.twoplustwo.com/25/probability/probability-set-over-set-380320/

Spadbidder is wrong when he says the odds of two people having a pair at a ten handed table is 9.8%. It is closer to 25%. Also the "two people seeing a flop is 24%" is very seriously flawed when you are talking about two people with pocket pairs. Much more likely that they will take a flop, obviously not 100% tho

Loverboy,

How about you test your hypothesis and then let us know the results.

While we're on the subject of probability and stats, here is a question. With over 12,000 posts in the "poker is rigged" debate thread at 2p2 and with over 80 billion hands dealt by the major online poker sites, what is the probability that all-in situations are rigged AND that no one has been able to produce a shred of credible evidence that this is the case? Seems infinitesimally small to me.

Loverboy,

How about you test your hypothesis and then let us know the results.

Yeah, you are right.

But three SD's is 3 SD's. It is commonly accepted that 5 SD's does not happen in a normal distribution.

13 samples isn't any kind of distribution, it's a smattering of data points from which no significant conclusion can be drawn, even if it was 1 mirrion SDs from the norm.

13 samples isn't any kind of distribution, it's a smattering of data points from which no significant conclusion can be drawn, even if it was 1 mirrion SDs from the norm.

You don't know what you are talking about. It is math, not opinion. Don't just say "I don't believe it!" either say "X+Y = Z proves that your findings are inconclusive" or don't say anything at all. Your opinion is totally uninformed, either inform yourself or don't spread misinformation. I haven't said one single opinion in any of my posts. No anecdotes, nothing like that. I agree that my sample size is too small but not by nearly as much as your intuition suggests. If you disagree with that statement, then support your thoughts with facts please. It should converge pretty quickly.
http://en.wikipedia.org/wiki/Checking_if_a_coin_is_fair

Sample size makes your confidence level higher and your confidence interval smaller.
This tells me that with a sample size of 13, yes a miniscule 13, I can be 90 percent sure that the actual percentage falls within the interval 0-38% , but at a confidence level of 99%, the confidence interval is 0-51.1% so I do need more datapoints.

he's saying you might be flopping more 2p hands than you should on average or something like that.

and as far as knowing if you are making good decisions or not; that is my entire point, the EV line doesn't say whether or not you played the hand correctly. the only way to know that is to review the hand.

for example lets say we raise a button, get 3-bet, and 4-bet with 99 intending to call a shove. he ships and shows AK and we lose the flip. you might say, well, i just got pwned in AIEV, but if your opponent is only getting it in with something like JJ/AK and you mistakenly think he's lighter, you're making a huge mistake vs his range that isn't accounted for in the EV line. in other words, the EV line only measures sklansky bucks, not g-bucks. in other words it only tells you if your EV vs his actual hand, and not his range. and since we know how much variance is involved with just the actual hands we are dealt, think about how much variance you're going to experience with running into different parts of your opponents ranges. this all isn't considered in EV lines.

furthermore, if you think your opponents range is wide in that scenario, it may be more profitable to call since we have position. all of these things are totally ignored by players that focus too much on EV line and not evaluating the actual merits of all their plays.

you really should review any pot you play with bets on multiple streets, as well as periodically review your c-bets, 3-bets, isoraises, etc. to make sure you aren't spewing.

Good input. It reminds me of an interview with Phil Galfond in which he said that Durr came up with a mindblowing play in a high stakes hand analysis on the 2+2 forum. A lot of players were debating Fold VS Call on the river and Galfond was a little more inclined to call when Durr said that shoving was the better play. It hadn't even entered Galfonds mind that shoving was a viable play there, to him it was mind blowing. Tell me what poker program could figure that out

you got a link to this interview/hand analysis?