Prop bet vs the "poker is rigged" believers

[-moe-]

This may be the wrong forum, but I couldn't find a better one. Sorry if I misposted.

Anyway, I think I've got an interesting case for a prop bet coming up, and I need some pointers about how to work out the probabilities and questions about statistical significance.

Context: a growing number of people in a regular home game I play in are convinced that online poker is rigged in various ways. I'm getting a bit tired of trying to argue with them, so I thought it would be just smarter to try to make some money off this, through a prop bet. Perhaps some of the smarter guys would be educated through the process aswell (I don't mind if they get better at poker, we're just playing for what amounts to peanuts anyway, and they are my friends.)

Now, the most common complaint lately has been that on Pokerstars there is always an "action twist" on the river card (thereby "Riverstars", obviously). The guys championing this idea, says that they are usually, and way more often than what would be expected by chance and compared to other poker sites, able to guess the river card before it drops.

As an example, if one guy was behind on the flop, then draws out on the turn, the river will way more often than what should be expected cause a re-suckout -- "on Pokerstars only", "a Pokerstars special" &c.

What I'm planning to suggest for the prop bet is therefore the following: shown a number of hand histories from Pokerstars where there is an all-in situation before the river, they will attempt to guess the river card. I take note of their guess(es) -- they may bet a number of river cards if they like, or they can pass that specific hand if they don't find a satisfying way there can be an "action twist".

I then present the list of actual outcomes, and with it go through their guesses and use a poker odds calculator to figure out the accumulated odds of the hits and misses, run this through some statistical analysis and gets some figure indicating what the accumulated chance of getting that amount of correct guesses is.

If they guess better than expectation -- and within a margin of statistical significance -- they win the bet. If not, ie they can not guess significantly better than what you would get by random chance, I win the bet.

What do you guys think, does this sound like a decent idea?

I expect working out the part about accumulated odds will take some work, figuring out how correct they need to be for it to have statistical significance, so I plan to make the bet as expensive as I can while still acceptable to them. Nothing less than USD $1000 (our buy-in at the home game is typically in the $50-$100 range, so we're by no means professional gamblers, so I don't think I'll be able to push it higher than that, even if the "omg rigged!" guys pools together resources).

I'm looking for any input you have on this idea.

If anyone could for instance give me some pointers about how to work out the statistics for finding the accumulated "correctness" of their answer, that would've been much appreciated.

How would I go about to make an estimate about how many hand histories I would have to give them to go through to meet some criteria for statistical significance (within two standard deviations? a P<0.5?)? Again, any input would be much appreciated.

(My plan here is BTW to play a bunch of turbo MTTs on Pokerstars, taking note of the tournament ids so they can double check themselves after concluding the bet that I didn't scam them.

Then go through the HHs, picking all relevant hands, anonymizing screennames so they cannot figure out results of earlier hands from later chip-stacks, snipping away information about what happens at the river, then print out the whole thing and give it to them next time we play at the home game. Then let them go through and suggest cards that they would expect to fall at the river, then go through that and run the calculations.)

[pzhon]

[ QUOTE ]
What do you guys think, does this sound like a decent idea?

[/ QUOTE ]

That sounds far too complicated and tedious. If they would follow that procedure, they would almost certainly not believe PokerStars is rigged. Why not just ask how often they think flushes hit on the river when all of the money goes in on the flop, and play a few cheap tournaments in front of them? Bet at a value between the fair amount amd their suggestion. You can also let them try to guess the river card, paying something between the fair amount and the frequency they say they can predict the river, say paying 40:1 if they think they can guess the river card 1 time in 30.

The down side is that you won't be a lock to win. However, you'll get more action, and it will be more fun.

[99killed]

tl;dr---we all know stars is rigged. u lose.

[-moe-]

[ QUOTE ]
[ QUOTE ]
What do you guys think, does this sound like a decent idea?

[/ QUOTE ]

That sounds far too complicated and tedious.

[/ QUOTE ]

Yes, but -- that's for me, not for them. They just get a HH snipped off before the river, and will guess what the river is. We'll just repeat it as long as they wish.

That's all, basically. Then I would have to do the work of a) providing hand histories, and b) figuring out the result after they are done guessing at as many hands as they like.

The main reason I want to do it like this, is that it tests exactly what they say they can do, so there can be no good excuses if (when) they lose the bet.

[ QUOTE ]
Why not just ask how often they think flushes hit on the river when all of the money goes in on the flop, and play a few cheap tournaments in front of them?

[/ QUOTE ]

I like the idea of betting on a per-hand basis. That would be quick and painless for me aswell. Thanks for the idea.

[ QUOTE ]
You can also let them try to guess the river card, paying something between the fair amount and the frequency they say they can predict the river, say paying 40:1 if they think they can guess the river card 1 time in 30.

[/ QUOTE ]

I like this one better than fixed at eg a rivered flush, since this would provide the possibility of a bet for a lot more hands. No need to ask them how often they expect to hit, though -- I can just calculate the actual odds in each case, which is dead simple and very quick. ("The river will be a 5 or a club!" "Ok, let's see, that's 11 cards, and there are now 44 unknown cards, that could come from the deck. I give you 3:1." :-))

[ QUOTE ]
The down side is that you won't be a lock to win. However, you'll get more action, and it will be more fun.

[/ QUOTE ]

Yep, thanks for your input.

[schwza]

if these guys are really that suspicious they'll think you grabbed a bunch of hands where there wasn't an exciting river.

[tonybormin]

Don't waste your time and just take their money in a cash game.

[-moe-]

[ QUOTE ]
if these guys are really that suspicious they'll think you grabbed a bunch of hands where there wasn't an exciting river.

[/ QUOTE ]

That one's easy to overcome. I can just store the whole HHs for them to inspect later, if they should accuse me of cheating. Would be extremely tedious and timeconsuming to fake a full HH, as you would have to do a lot of "tuning" to doctor hands while still keeping the chip-counts around the table in sync.

Anyway, these are friends, and I don't think they would distrust me. They just distrust Pokerstars (and other sites), based on some psychological phenomenon (false pattern recognition, confirmation bias, whatever).

[-moe-]

[ QUOTE ]
Don't waste your time and just take their money in a cash game.

[/ QUOTE ]

I already am.

I'm now looking for a way to convince them that Pokerstars does not rig up how the cards fall in an all-in, so I can simultaneously enjoy playing with them while I am winning their money.
[img]/images/graemlins/smirk.gif[/img]

(All the talk of rigging is getting to be a bit too much.)

[Sandwich]

I'm having trouble seeing how this will work practically.
These guys, who are asserting that Pokerstars more often than not deals suckouts on the river are supposed to GUESS what the river card will be? Seems asinine. Are we expecting them to guess suckout 100% of the time, or they have to guess which ones will be suckouts and which will not?

Wouldn't a better way of rebutting them be to have THEM pick a sample of n hands from hand histories and bet them a large amount of money that the hands, grouped by percentages, stay true to their percentages (within a certain confidence interval) over that n number of hands?

For example, 70-30s will get sucked out on 30% of the time... not more, as these guys would argue.
60-40s will get sucked out on 40% of the time... etc. etc.
?

[-moe-]

[ QUOTE ]
Wouldn't a better way of rebutting them be to have THEM pick a sample of n hands from hand histories and bet them a large amount of money that the hands, grouped by percentages, stay true to their percentages (within a certain confidence interval) over that n number of hands?

[/ QUOTE ]

Umm... this is exactly what I suggested in my original post.

Perhaps I was too verbose and not expressing myself clear enough.