Facts and Myths 3, Fun With Wheels

The "Breaking Vegas" show had two eps about wheels. One was about biased wheels, the toher about using the laws of physics to rpedict where the abll would land.

Biased first. The theory is over time wheels wear out in different ways, they tilt minutely to one side, say, or become slightly off-balance. In any case by plotting the frequency of each number over thousands of turns of the wheel, one can determine which numbers turn up more frequently in a given wheel.

It sounds reasonable, though also like a lot of work. Keeping track of numbers is easy, as many csainos even provide pencil and paper at roulette tables. The show mentioned tracking about 10,000 numbers. That's several days' worth at least. There were no specifics on whether the tracking was done continously all day long or across shorter sampling sessions.

The reason I say it sounds reasonable is that we all know other artifacts that show some form of minor malfunction that could be called a bias. A toaster that burns a little too much, a car that pulls slightly left, an office chair that can't roll straight backwards, a mouse with a balky left button, etc. So why not roulette wheels? some wear on the mechanism would cause a kind of bias, perhaps.

I'm undecided on this matter.

The other way of beating the house is this:

Using the laws of mechanics and measuring the speeds of the ball and wheel, you can make a fairly acucrate prediction of where the ball will land. Not down to a number, but to a section of the wheel. You divide the wheel in 8 parts and thus can tell in which octant the ball is likely to land, and bet on those numbers.

You'd know some non-variable data, too: the radius of the wheel, the ditribution of the numbers and the rules for laying bets. Now, this is where I see apossible objection: can you place bets while the wheel and ball are spinning? I've never played roulette at a casino. I'm assuming you can, because otherwise the whole myth fails.

Next, the calculations are complex. You can't do them in your head, at least not in time to make a bet. Therefore you need a computer. That is against casino rules and possibly illegal. You also need to measure the speeds cited above and input them into the computer. Lastly you need to be able to read the computer's output. These complications are serious since the computer, input and output devices have to be hidden from the casino.

I'm convinced the system would work, as far as giving you the ability to predict the likely octant for the ball to land on. The ball does obey the laws of physics, after all. Whether it can be implemented practically, with enough time to place the necessary bets, is a different question.

According to the TV show it didn't quite work, precisely as regards th practical implementation. this was in the 70s, though. Today it would be easier using blue tooth links for input and output, and you might even be able to load and run the prediction algorythm in a cell phone. Nothing unusual or suspicious about carryign a cell phone and a blue tooth earpiece, not these days (maybe a bit in the casino, but make a few calls and presto).

This, however, is cheating, as the rues do not allow electronic devices to aid you in the game. It may even be illegal.

A bit of trivia, the same system was used in an episode of the origins "Mission: Impossible" TV series. The objective was to break the bank of a European arms dealer's casino, making him unable to pay off his suppliers (there was also some cheating at Baccarat). The ending had a very funny line:

"This should teach him money can't buy love. In fact, sometimes it can't even buy guns."

The system also made an appearance in CSI. The perpetrators wound up dead, but not killed by the mob-tied casino owner.

Quote: Nareed

The "Breaking Vegas" show had two eps about wheels. One was about biased wheels

I liked those shows too. As far as the biased wheel, it was remarked the edge it gave knowing the bias was good on European wheels, but not U.S. wheels.

I was really surprised there were such things going on.

I must have missed the rest of the show, don't remember trying to use laws of physics on un-biased wheels.

Roulette:

Check out the results of Billy Walters at the Golden Nugget in Atlantic City, 1986, to the tune of 3.8 million by betting $2k straight up on each of 7-10-20-27-36. That should convince you pretty squarely that Roulette wheels have biases that can be exploited. Yes, that was 23 years ago, yes it is still possible today. It is a big planet.

Quote: DorothyGale

Roulette:

Check out the results of Billy Walters at the Golden Nugget in Atlantic City, 1986, to the tune of 3.8 million by betting $2k straight up on each of 7-10-20-27-36. That should convince you pretty squarely that Roulette wheels have biases that can be exploited. Yes, that was 23 years ago, yes it is still possible today. It is a big planet.

You've got me looking that up!

Billy Walters at the Golden Nugget

With respect to bias,

Absoultely, roulette wheels can probably have a bit of bias where over time, certain numbers would have bias over others. The way to analyze this would be to only analyze all of the numbers hit over time and look at the frequency of hits on a roulette wheel and do your statistical calculations to find the section of the wheel that hits most and then bet on that number. But even still, you have to overcome the 5.26% house advantage meaning that you need to ensure that these numbers will hit 1/36th of the time or better.

Let's say for example that you find a sector of 5 numbers that are more prevalent. Over 8,208 spins (36 * 38 * 6), the expected number of hits on those six numbers is 1,080. The variance can be expressed as (np(1-p)) ^ 1/2 = (8,208 * (6/38) * (32/38))^ 1/2 = 30.62. Two sigmas out generally means that anything outside of that variance is only 5% likely to occur and represents a true bias. So, if you observed 1,141 hits or more on those sets of five numbers, you would be seeing a likely bias. It so happens that the 1,141 hits out of 8,208 represents an occurrence of 2.78% per number for a player advantage of 0.0877%.

So as a methodology I would record a pile of hits and simply look for variance trends over sets of numbers and bet accordingly if possible. Note that there is no point on looking at one number only because of the variance in the ball itself which will affect the result.

--
With respect to a dealer influence on the ball, you can bet while the wheel is spinning to the point where the dealer states that there no are no more bets. I think all you really care about is the release point of the ball and the end result and measuring the difference. For example, if the ball is released in sector A and lands in sector D beyond a variance as noted above, you can take advantage of that. But by the time you discovered an actual verifiable variance, it would be late in the dealer's shift!

Casino roulette operators have been trained to release the balls at different speeds since that flaw was discovered. Still you might find someone who tends not to follow that regime.

The typical way to approach roulette bias is to use chi-squared testing based on 10k spins or more and look for terms that have:

(Obs - Exp)^2 / Exp > 2.

This is what Bill Zender descirbes as an acceptable methodology in "Advantage Play for the Casino Executive."

--Dorothy

-----------------------------------------
DG's minimally offensive signature

>European wheels, but not U.S. wheels.

just to clarify, the edge is slight and the less generous U.S. wheels don't cut it according to the program I saw.

Billy Walters at the Golden Nugget must have found the most biased wheel imaginable to beat the 0/00 wheel it must have been. Assuming that is in fact why he won, and the story is factual.