[dageroth]

I bootstrapped my first startup by playing roulette in a casino every night. A Cinema close by was giving out 10 Euro coupons for the casino to play with, so we went first to the cinema every night, asked the people coming out for their tickets and then hit the casino to play with the 10 Euros. Each of us were netting about 400 Euros per month, which was enough to cover the ramen ;-) And that for a nice refreshing half an hour tour in the evening.

The stupid EU put an end to it by changing the gambling laws and prohibiting casinos to give out coupons for chips - someone might get addicted - instead the new coupons allowed free entry and a drink, but until then it worked like a charm.

The casino people were a little freaked out at first, shooting us suspicious looks but after three weeks they got used to us...;-)

[nollidge]

Isn't roulette a losing game? Or did you have some optimal betting strategy?

[dageroth]

We didnot play with our own money, but with the coupons on the back of the cinema tickets. But even so, we should have made slightly less than 300 a month, but we did a slightly more on average.

We did not really have an optimal betting strategy, at least nothing that can be proven scientifically, there is no such thing. But after some time we started to bet on the wheel, e.g. "Voisins du zero" or "Jeu zero" and we knew all the croupiers and some of them had an unbelievable regular way of running the ball and spinning the wheel and the ball would thus drop in a slightly more predictable area of the wheel - at least when it hit one of the bumpers and would then drop into the wheel. So we waited with our bet until our favourite croupier would throw the ball from a place, where we felt it should land close to the zero and call a zero game.

Naturally that did not always work and we'd have losing streaks of even two weeks where we went home without any money at all, but once a month or so we'd go out with over a hundredfifty euros. On average we did slightly better than expected (400 were admittedly the better months, sometimes it was a little less than three hundred), but we remained ahead even after a long time, so my guess is we were either lucky or the idea of predicting according to the croupiers regularity worked. We had the feeling it worked on two of the ten croupiers the others threw less regular.

Anyway it was definitely more fun than just whitewashing the chips. (One was not allowed to immediately exchange the "lucky" chips one would get for the coupons for money.)

[settrans]

> We did not really have an optimal betting strategy, at least nothing that can be proven scientifically, there is no such thing.

Ah, but there is.

http://en.wikipedia.org/wiki/Kelly_criterion

[zb]

The optimal (Kelly) strategy for Roulette is not to play. It even says so on that wiki page.

[rfrey]

The Kelly criterion is a money management strategy to maximize profit given a positive expectation, which roulette does not have and cannot.

[jdp23]

Well done! This is the no-tech version of what the shoe computers were trying to do ... with a lot fewer shocks :-)

[metamemetics]

Roulette can be beaten by a shoe computer. Basically you tap your foot each time the ball passes a certain number. The computer uses the time between taps to compute the velocity of the ball, which is used to predict where it will stop. Then you bet on a quadrant of the wheel by the predicted stop.

This was first done by people from MIT I believe.

[metamemetics]

Excuse me, UCSC students in the 1970's. Those interested can read more here: http://physics.ucsc.edu/people/eudaemons/eudaemons.html

[lesterbuck]

The book about that group, The Eudaemonic Pie, is a great read:

http://www.amazon.com/Eudaemonic-Pie-Thomas-Bass/dp/05951423...

Recently I ran across another book about the same group, "The Predictors: How a Band of Maverick Physicists Used Chaos Theory to Trade Their Way to a Fortune on Wall Street"

[tlb]

A poorly written book about an interesting topic, I thought.

[nailer]

You'd no longer have the opportunity to place your bet by that time.

[dageroth]

that's quite brilliant, it should work probably much more reliable than our guessing.

[tb]

The way I read it, they were betting with free money, so they couldn't lose.

[gcheong]

Essentially they were converting the euro coupons which they got for free to cash with the expected value of a bet on the roulette wheel (which is slightly negative giving the house an edge over time). For example, if the avergage expected value of the bet is -5% (the house edge), then for a single bet, on average, you would expect to get 95% back of your original bet amount. Since they got 10 euros each time for free they would expect to get back 9.5 euros on average over all their bets, given this house edge (which I made up).

[fragmede]

It's not a losing game if you don't use your money to play (the coupons dageroth referred to), and you get to keep any money you win. (The 'hack' is you weren't supposed have ~€800 worth of coupons.)

[viggity]

The house edge for a Red/Black or Even/Odd bet in roulette is about 5%. Meaning that if he on average walked out with €400, you would expect that he walked in with €420 in coupons