hey guys,
i am opening this thread because of a very important bet i have just made with my very smart boyfriend, because he didnt believe me the following theory about roulette:

I read in my psychology studies that when you bet on the SAME color each time you will lose less money on the long run than if you would switch it up (as in following your "gut feeling").




Karen

Hi Karen,

Either your psychology instructor sucks at probability or your misunderstood him/her. The expected value of betting red or black on a single spin in roulette is the same no matter what you bet (assuming the wheel is truly random). Your boyfriend wins this bet.

However, your idea is not totally off base. If someone tells you that in 100 games, team A will win 70 times and team B will win 30 times and you have no other information about those games, you are better off just betting on team A every time. Many people intuitively try to "probability match" by betting on A 70 times and B 30 times. While this is the only strategy for getting all 100 games perfectly right, on average it will do worse than simply betting on team A every time (in which you are guaranteed to be right 70 times and wrong 30 times).

http://en.wikipedia.org/wiki/Indepen...bility_theory)

+1

That's like saying it's more profitable to choose heads 100 times in a row when having a coin flip marathon. It's not.

Perhaps people bet less on average when they do not bounce between different bets based on their gut (chasing losses less for instance). That would explain why they lose less - it would be a function of how much a person bets, so the psychology aspect could be along those lines much like how when casinos started showing the past numbers the revenue went up because people bet based on patterns.

The OP left out human behavioral tendencies as part of the equation (ie: whether it is part or not).

Bet with him.


Unless you're making all 100 bets in advance, you'd do better by betting on B sometimes. For example, I'd always get the last game right, and that means that there is a 30% chance that I'll be betting on B in the last game (assuming 70/30).

But in Roulette where the spins are independent, and the probability of red and black are equal, it makes no difference which you bet on. Your probability of winning or losing any given amount after any given number of spins will be exactly the same, and on average you will lose 5.26% of the total money that you bet if it's a 00 wheel.

Yes Bruce, I was presuming that you had to lay all 100 bets in advance.

To use a poker related situation, consider playing against a particular player who makes a particular bet with no more bets possible (either all in or on the last round of betting). You know that he would only do this as a bluff (30%) or with the nuts (70%). You are getting the right price to call because he does it as a bluff often enough. How often should you call? Of course the answer if every time you are in this situation. You should not try to guess which situations are the bluffs and which situations are the nuts. Obviously if you have some sort of tell that let's you know otherwise, you can do better than a pure "always call" strategy, but in the absence of such extra information always calling does better than trying to guess which times are bluffs and which times are not.

Please describe where you read this, and if it's really in some kind of published scholarly material you should complain loudly to the idiots who wrote this.

Likely the material had nothing to do with the odds of the game but rather the conditions in which people tend to bet larger amounts. Casinos always try to make their games more amenable to average bettors betting more, so I suspect this is tied more into that.

Was there anything in the article about "PSI missing"?

obv is es latte. jedes mal is es wieder neue 50%. spielt 0 rolle was du wann nimmst. es muß jedes mal random sein!
weil das universim des letzten males existiert nicht mehr. es ist jedes mal (ein) "neues" universum/neue odds. history hat nur im metagame relvanz
stochastisch gesehen gibt es keine History.
Nur zwischen und für die/denen die sich an das alte uuniversum erinnern
gibt es history, aber das universum/unsere dimension hat doch kein gedächtnis;
und es hat doch immer noch infinite zeit um wieder breakeven zu kommen...

can somebody proof it, somehow?

By definition, random events in the past can't affect future random events. Otherwise they would not be random.

Put another way, every spin of the roulette wheel is the first spin.

I don't understand why D:

Team A wins 70 and team B wins 30. If you bet team A every time, you win 70 times. If you bet Team A 70 times and team B 30 times, on average you win 70% of the 70 games that you bet A, and 30% of the 30 games that you bet B, for an average of 0.7*70 + 0.3*30 = 58 games.

oh, I totally misread that. Wait people actually bet 70 times on A and 30 times on B in that scenario?

Yeah, it didn't help that his numbers didn't add up to 100. Sherman, I took the liberty of fixing your post (A wins 70 and B wins 30, not A wins 70 and B wins 70).

That is what I meant to type, so I am glad you fixed it.

Oops! Hope you didn't bet too much!
Think critically about everything you read and hear. Especially question a claim made about something not in the author's area of expertise (I wouldn't go to a psychologist to learn about mathematics, nor to a mathematician to cure my craziness). But I think Monteroy's explanation is likely, and you didn't read carefully enough. Maybe you only saw the data and only thought of 1 possible interpretation (which happened to be the wrong one).

I'm a psychologist and people come to me all the time for statistical questions. Statistics and research methods is pretty much a requirement for any BA/BS in psychology in US (exceptions may apply, but I personally would suggest transferring to another university to get a psych degree than to get one without these topics covered). My minor area of study for my Ph.D. was in quantitative psychology. That being said, psychology is much more about applied statistics than about probability theory / mathematics (which no doubt explains my frequent mistakes on this forum about combinations and pure probability problems).

For what it is worth, I doubt this is something that was read in a psychology class. I think the most likely explanation is that either 1) the student remembered what the instructor said incorrectly or 2) the instructor got it wrong. The reason for this that most students don't read (my students always find it amazing when I tell them about a neat study that was in the book; if they were reading they should be bored by it).

Further, despite (research) psychologists have reasonable training in quantitative issues, many of them still suck at it, which makes #1 plausible. #2 is also plausible because...well, see the scores on my exams for the evidence against student memory. I'd rate "didn't read carefully enough" as the third possibility and "book error" as an outside 4th.

So in order of likeliness:
1) Student misheard/misunderstood/misremembered the lecture
2) Instructor said it incorrectly
3) Student misread/misremembered book
4) Book is in error.

My statement wasn't meant as a knock on psychologists' math skills, though it was superfluous to my point. We should question a math claim made by a mathematician just as much as we should question one from a psychologist. I was just saying that if we're gonna be lazy about our critical thinking, we should spend more effort questioning a cross-field claim in general. But I guess the roulette example isn't really cross-field because there is a lot of statistics in psychology as you've pointed out.

This is possibly something to do with confusion surrounding this now widely debunked study.

It could also be to do with the fact that players chasing their losses at roulette tend to move from predictable outside bets (eg most players stick to black or even etc) to unpredictable inside ones (where they just splash the chips everywhere around certain numbers).

This creates a vague correlation in casino data between a higher hourly loss rate and unpredictability in betting patterns, although I doubt this would have made it into much of the general psychology literature.

This 'extra' loss also simply comes from a higher average stake when people are steaming, a higher average stake which happens to coincide with a general change in betting style, not any other strange factor.

This is a classic argument in my opinion and I always like to bring up this point;
Lets say we were flipping a coin ten times.
For the first five flips to our surprise comes heads.
Now for the following flips ; 6,7,8,9,and finally 10.
Probability says the results are individual but theoretically if I bet
3x on tails the 6th bet , 9x on tails the seventh bet, and increasing the bet size(12x,25x,65x) on
tails each time heads wins; I should come out ahead.
(also we decided not to bet on the first five flips.)

No, Martingale-type betting (or any other type) does not change the expectation of the game, your point is false.

Good god you will get crushed. Yes, most of the time you will win a small amount. But you WILL hit a long enough streak (and much sooner than you probably think) in which you will lose more than you can pay. As NewOldGuy says, look up Martingale on wikipedia.

I'm quite sure you are "wrong" about the outcome of this bet. It is the girlfriend that wins. In bets between boyfriend/girlfriend or husband/wife, the normal EV equations do not apply. The believe following scenario illustrates the real situation accurately:

If a man expresses his opinion in the middle of the forest and no woman is around to hear it, would he still be wrong?