Prisoner's dilemma and syndicate betting (Video)

[Justin7]

[mathdotcom]

It's funny how you setup the prisoner's dilemma and then say almost nothing about why this syndicate has suddenly been able to cooperate mid-season. If no one knows who is betting the games, you can't identify who exactly is cheating. So grim punishment strategies, etc., will not work. The outcome will always involve everyone cheating.

By the way, here's another prisoner's dilemma for you, Justin

Players: Justin7, VegasDave

Players are roommates.

Justin7/VegasDave Leave underwear around house----- Don't

Leave underwear ------0/0--------------------------------- 3/-1
Don't -------------------1/3--------------------------------- 2/2

Both players obtain utility from leaving their underwear lying around house, but get significant disutility from having their roommate leave underwear around the house. Neither can resist temptation to leave underwear around house since they know other roommate will do so as well.

Perhaps this issue could be resolved by Justin7 banging Jennifer Love Hewitt on Vegas Dave's bed?

[Justin7]

Sorry, Math. The vid was already too long. If you want to know how it happened, check out two sources:
Game theory and Strategy by Philip Straffin, and a Science News Article whose cite I cannot find.

[RogueScholar]

A few issues with this video:

1) The term cooperate is not just ambiguous, but in this instance it has contradictory definitions. Most people (myself included) that have run afoul of the law and spent time in the pokey understand the word cooperate in this context to mean cooperation with the police/authorities. Since you're never allowed to communicate with your cohort when in custody and facing questioning, you can't really cooperate with each other in the traditional sense, beyond a committment not to screw over your poor buddy who is also under the bright lights. You use the term to indicate cooperation between the prisoners, which while quickly evident to those who have studied this particular wrinkle in game theory, might not be immediately clear otherwise. I felt this deserved some amplification.

2) I would caution those to have watched this video not to follow this advice blindly should they encounter a real-life iteration of it. If you and your buddy get picked up for doing something shady, and he remains in jail while you walk out later, the fact that you're only facing the dilemma once doesn't necessarily make being the first one to be a "rat" a +EV decision.

Your buddy may have friends on the outside who won't take too kindly to even the perception that you rolled on him. Perhaps you would find yourself in a dark alley with some scary guys who bust a cap in your ass not long after your release. In that scenario you would have been better off sitting in jail, playing tonk and making pruno than being a rat. Using the toilet with only a towel over your lap for privacy and smoking cigarettes filled with tobacco smuggled into jail in someone's rectum sure beats being dead. The moral of the story? Being a rat is never...ever...+EV.

3) You infer in your closing that these syndicates are in collusion with each other to guarantee profit for all. I find it hard to believe that all the world's sharps are so uniformly cooperative. Most people I know are given to fits of greed, and while I'm sure there's a "brotherhood" of people making money at betting sports, the law of averages tells us that in the group are gonna be some real assholes. How exactly do you keep the assholes in line?

I think there's a greater sociological lesson to be learned here, and I think were I to possess it, I might in fact be able to rule this forum with an iron fist. I promise you that my autocracy will be far more elitist than the current administration, and that you will be revered as a cult figure by my loyal subjects should you share this information with me. Unless of course you rat me out to Willie Bee, which results in the dark alley scenario, which is decidedly -EV. I think we could really change the world here, please give it some thought.

Is there any way to join this syndicate or consortium of syndicates? Do any of you guys smoke weed? Have needs for oral sex that aren't being met? I'm a man of unusual resources, please don't dismiss my sincere interest in joining your ranks out of hand. I can sing, dance, and play the piano too (although not all at the same time).

Thanks for the video, not only was it entertaining, but it was the impetus for the first fun post I've written in months. See you on the 18th!

[mathdotcom]

Justin, you can't answer it?

This is an N-player game where cheaters cannot be identified, not the simple 2x2 example you provided. The "cheating" equilibrium cannot be avoided.

[InTheHole]

Very good Video

[mathdotcom]

A few issues with this video:

1) The term cooperate is not just ambiguous, but in this instance it has contradictory definitions. Most people (myself included) that have run afoul of the law and spent time in the pokey understand the word cooperate in this context to mean cooperation with the police/authorities. Since you're never allowed to communicate with your cohort when in custody and facing questioning, you can't really cooperate with each other in the traditional sense, beyond a committment not to screw over your poor buddy who is also under the bright lights. You use the term to indicate cooperation between the prisoners, which while quickly evident to those who have studied this particular wrinkle in game theory, might not be immediately clear otherwise. I felt this deserved some amplification.

2) I would caution those to have watched this video not to follow this advice blindly should they encounter a real-life iteration of it. If you and your buddy get picked up for doing something shady, and he remains in jail while you walk out later, the fact that you're only facing the dilemma once doesn't necessarily make being the first one to be a "rat" a +EV decision.

Your buddy may have friends on the outside who won't take too kindly to even the perception that you rolled on him. Perhaps you would find yourself in a dark alley with some scary guys who bust a cap in your ass not long after your release. In that scenario you would have been better off sitting in jail, playing tonk and making pruno than being a rat. Using the toilet with only a towel over your lap for privacy and smoking cigarettes filled with tobacco smuggled into jail in someone's rectum sure beats being dead. The moral of the story? Being a rat is never...ever...+EV.

3) You infer in your closing that these syndicates are in collusion with each other to guarantee profit for all. I find it hard to believe that all the world's sharps are so uniformly cooperative. Most people I know are given to fits of greed, and while I'm sure there's a "brotherhood" of people making money at betting sports, the law of averages tells us that in the group are gonna be some real assholes. How exactly do you keep the assholes in line?

I think there's a greater sociological lesson to be learned here, and I think were I to possess it, I might in fact be able to rule this forum with an iron fist. I promise you that my autocracy will be far more elitist than the current administration, and that you will be revered as a cult figure by my loyal subjects should you share this information with me. Unless of course you rat me out to Willie Bee, which results in the dark alley scenario, which is decidedly -EV. I think we could really change the world here, please give it some thought.

Is there any way to join this syndicate or consortium of syndicates? Do any of you guys smoke weed? Have needs for oral sex that aren't being met? I'm a man of unusual resources, please don't dismiss my sincere interest in joining your ranks out of hand. I can sing, dance, and play the piano too (although not all at the same time).

Thanks for the video, not only was it entertaining, but it was the impetus for the first fun post I've written in months. See you on the 18th!

1] C'mon, give him a break, there are 100 different names for the strategies. "Rat/Clam", "Cheat/Dont Cheat", "Cooperate/Dont Cooperate".

2] That's not the game analyzed. There is no assumption of mafia, etc. If cheating means you get killed when you get out of prison, then it's not a prisoner's dilemma because the payoffs are all different.

3] At least the first, non-anal part of this, is part of what I asked Justin. Too many players and cheaters can't be identified.

[Justin7]

Justin, you can't answer it?

This is an N-player game where cheaters cannot be identified, not the simple 2x2 example you provided. The "cheating" equilibrium cannot be avoided.

You're obviously correct. Assume there are N syndicates. In the first game, someone cheats. In the following game, the other N-1 syndicates execute a "punish" strategy. Before full overnight limits of $2k are reached, punish is implemented at $500 when the first book comes out.

Third game: the players watch. If someone cheats, a punish strategy is again implemented. I'm mad, because I cannot find the article on this... But in an experiment with 100's of gaming theorists, the strategy that did best was "cooperate, punish if they cheat, and offer cooperate again".

Look at WNBA markets. Cooperation has been achieved. Last year, I led a co-op effort in Arena as well (and had my first 6-figure year in it).

[mathdotcom]

Justin, search 'tit for tat' and 'grim punishment strategies'. That's what you're thinking about; these are used in a field of economics called 'experimental economics' all the time.

Your analysis is wrong. There's not just one cheater and everyone else is trying to find him. Everyone knows that everyone else will cheat because the cheater can never be identified. The cheater never stops cheating, despite punishment, because he knows if he stops cheating someone else will.

[mathdotcom]

Your theory about WNBA is just one theory. It could be that the linemakers got a clue and stopped putting up bad lines, among other theories.

I'm questioning your theoretical analysis. I am not commenting on whether a syndicate can be successful.

[RogueScholar]

1] C'mon, give him a break, there are 100 different names for the strategies. "Rat/Clam", "Cheat/Dont Cheat", "Cooperate/Dont Cooperate".

2] That's not the game analyzed. There is no assumption of mafia, etc. If cheating means you get killed when you get out of prison, then it's not a prisoner's dilemma because the payoffs are all different.

3] At least the first, non-anal part of this, is part of what I asked Justin. Too many players and cheaters can't be identified.

1) I wasn't saying that he was in error, I was simply pointing out that the number of posters on SBR familiar with the nomenclature of game theory might be a rather significant majority. My response was designed to clarify and to entertain. I meant no disrespect, the problem was described adequately.

2) I disagree. The parameters of the classic prisoner's dilemma end abruptly at the decision of the first prisoner to rat or the determination by the captors that neither is inclined to incriminate the other. After that it's all speculation and some researchers have indeed studied the possible scenarios after the conclusion of the initial outcome. I point you to this Science News article which doesn't address my response directly, but illustrates the possibility of accounting for variables after incarceration.

The Influence of Influence in Prisoner's Dilemma

3) I agree, I find it difficult to ascertain how a "punishment" could be exacted on the greedy. Absent an exchange environment where all parties are equal and anonymous, it strikes me as odd that the bookmaker(s) in question would just let these players settle the matter amongst themselves. They would seem to benefit from only a handful of the most greedy being achieving maximum leverage and certainly not over an indefinite period of time.

[mathdotcom]

1] I thought it was clear that 'cooperation' = 'cooperation between 2 prisoners by shutting up'

2] The prisoners decide simultaneously and are not in contact with each other. The game ends after both players make a decision. If you don't think it's realistic in certain situations, fine [mafia, chance you run into each other in line at Starbucks]. But those are the assumptions of the classic prisoner's dilemma.

[Justin7]

You both raise many good points. Why don't you do a response video addressing these, and limit it to 6 minutes?

[RogueScholar]

1] I thought it was clear that 'cooperation' = 'cooperation between 2 prisoners by shutting up'

Which actually proves my point, illustrating that you've never faced this scenario in real life. I'll tell you this much, from my own experience, when I hear a description of two co-conspirators being questioned in the hopes that one of them will rat, and the word 'cooperate' is used, my asshole becomes a size of a decimal point (not an insignificant feat for those who know me). You don't even have to know the people involved, your heart sinks knowing that the greedy asshole gets the +EV outcome and the honorable one gets to shit with a towel over his lap. It's just plain sad.

Mostly I'm beating this dead horse because I'm upset that you ruined my funny. This video was posted in Player's Talk, and my response was a Player's Talk response meant to make my fellow hooligans and misanthropes chuckle at my descriptive language and utter irreverance for the scholarly topic at hand. Now it's ruined.

The simpletons will lose interest halfway through the thread, thinking that it's some discussion pertaining to winning, which is patently not what we do in PT. We lose, and we do so with panache, drama, wailing and gnashing of teeth. In between, we make off-color remarks and funnies. That was the inspiration behind my initial post in this thread, but now it's too late. My parade has been rained on. My ship has sailed. My toilet won't flush.

I hope you're happy.

[mathdotcom]

Justin, we don't expect you to address everything in 6 minutes. Just come to the right conclusion with regards to this syndicate. I do not think it is applied correctly. If you said it's a prisoner's dilemma and that's why this WNBA syndicate did not work, then it'd be consistent with the dilemma you described.

[Justin7]

Justin, we don't expect you to address everything in 6 minutes. Just come to the right conclusion with regards to this syndicate. I do not think it is applied correctly. If you said it's a prisoner's dilemma and that's why this WNBA syndicate did not work, then it'd be consistent with the dilemma you described.

Look at the early line moves. Look at the game day line moves. Do you still draw the same conclusion?

[RogueScholar]

Indeed, I think we're simply lacking salient details to the conundrum of the rogue syndicates. Would it be possible to learn how you exact punishment on these greedy fuks? Mathy is right, the anonymity which is crucial to sports betting would seem to be a disincentive for anyone to stop cheating while simultaneously making it very difficult to "ride herd" on the cheaters. Perhaps there is something we're unaware of?

[RogueScholar]

Look at the early line moves. Look at the game day line moves. Do you still draw the same conclusion?

At which book(s)?

[mathdotcom]

Which actually proves my point, illustrating that you've never faced this scenario in real life.

My 'favorite' personal example is traveling on the train. Everyone stands by the doors because they either want to be the first ones off or are worried with all the other people on the train they won't be able to get off at their stop. But by grouping by the door they make it harder for everyone to get off and on, not to mention that they ride in discomfort. If they dispersed evenly then no one would have any trouble getting off and on.

[pico]

not really useful in gambling. i remember bunch of nobel prize winners tried to use math to gamble on options then us gov forced 40 investment banks to bail them out.

[mathdotcom]

Look at the early line moves. Look at the game day line moves. Do you still draw the same conclusion?

Your video is correct in everything you say - and it makes sense when applied to your story about why at the start of the season they weren't able to properly coordinate. But then you claim they fixed it mid-season because, I guess, they suddenly realized they were playing a prisoner's dilemma game and that everyone would be better off cooperating? In theoretical terms, the cooperation outcome is not a Nash equilibrium if a single player has an incentive to deviate from it. I'm claiming everyone does, because if they cheat they can't be identified and hence they cannot be punished. Simple.

In this quote I highlight here, you don't defend your argument which attempts to explain some event. You respond by asking me to solve it instead.

[Justin7]

Your video is correct in everything you say - and it makes sense when applied to your story about why at the start of the season they weren't able to properly coordinate. But then you claim they fixed it mid-season because, I guess, they suddenly realized they were playing a prisoner's dilemma game and that everyone would be better off cooperating? In theoretical terms, the cooperation outcome is not a Nash equilibrium if a single player has an incentive to deviate from it. I'm claiming everyone does, because if they cheat they can't be identified and hence they cannot be punished. Simple.

In this quote I highlight here, you don't defend your argument which attempts to explain some event. You respond by asking me to solve it instead.

A key point that many people don't realize... N is NOT a large number. The entry cost to getting good lines on WNBA is about 30k (cost of a scraper, math guy and programmer). If the total utility for a season is only 10 units, a lot of bigger players won't bother with it - their time is better spent elsewhere. My best guess is that N=4, and I am aware of 3 of them.

[mathdotcom]

A key point that many people don't realize... N is NOT a large number. The entry cost to getting good lines on WNBA is about 30k (cost of a scraper, math guy and programmer). If the total utility for a season is only 10 units, a lot of bigger players won't bother with it - their time is better spent elsewhere. My best guess is that N=4, and I am aware of 3 of them.

I'd like to coin a new SBR term:

Square by association

[Dark Horse]

[xyz]

I want to look at this from another angle. This would mean that the time interval when the books change from over night limits to the larger day of the game limits would contain many line moves in a short period of time across the market. Can anyone confirm this hypothesis? Thanks for your insight.

[Dark Horse]

The WNBA is a league of its own. I would like to know the market size compared to other sports. Seeing the fast adjustments (not just the one discussed here, but also in totals; which tightened up to a degree that is not seen in the NBA), I would be somewhat cautious. "After about a week there was discussion..." (near end of video). Unless that is a loose statement, which I doubt, this is a comment on the WNBA market. Who are making up this particular section of the betting public?

To approach a line movement topic from this specific angle (prisoners dilemma) could, possibly, be read as an invitation, a kind way of saying: "stop betting into the f*cking early lines".

Prisoners dilemma? Why not give it a little spice? Take the latest Batman movie. How many prisoners dilemmas sprinkled throughout this gamblers flick to send Gotham City into absolute Chaos? The bank robbery (kill all robbers), the Harvey/girl choice, the boats, the Batman-or-Joker theme. So what, in the long run, does work better in sports betting: chaos or order? (we'll leave that question to the Riddler).

I have to be honest. I don't see the resemblance between Justin and the Joker. This is more along John Nash lines. A beautiful mind. 'If everyone goes for the blond ...'

[durito]

greek doesn't even raise their wnba limits on gameday $500 sides $200 totals overnight and gameday

[Justin7]

you can get 1k if you call in on sides.

[Dark Horse]

Two NFL cappers each pick 58% ATS. Should they consider joining talents or are they better off going it alone?