Cyclical Luck

[Legato]

[ QUOTE ]
If he is talking about "The Eudaemonic Pie", it is considerably more complicated than that. They were a group of grad. students who modelled the physics and found that if you note the starting position and speed of the ball (often constant for a particular croupier), as well as some of the properties of that particular wheel (most are not perfectly balanced) you can predict with a reasonable edge which quarter it will land in. They had several observers and a computer/transmitter/reciever in a shoe. They claim to have made some money, but eventually stopped because the computer in those days was not reliable enough in the humid atmosphere of a shoe.

I am sure they made more money with their book,
Craig

[/ QUOTE ]

The phenomena still exists and works fine. Very expensive and no good to be seen by casino staff though...

There used to be pages called newtonroulette.com and roulettehardware.com but both are apparently removed. They used a system with a Ipaq in your pocket, a small earphone and a remote button to be held in your hand to register the speed of the wheel and ball. The earphone would then tell you an area where the ball would land which would be good enough to easily beat the casino edge.

Btw guys, if u got this great offer, how large percentage of your bankroll would you bet each time (you have to bet the same percentage every time)?

"Flip a coin 1000 times. Each time you chose side correctly you get 3 times your bet back".

[Copernicus]

[ QUOTE ]
On the other hand, for the puppies, the weights of the predicting factors tend to distribute themselves in such a way that that the degree of confidence in the prediction can't beat the rake either.



[/ QUOTE ]

As I noted above, this has been shown to be true for horses as well, by neural net and bayesian analysis.

[Cherenkov]

I noticed that many in this thread are concerned about "luck" and rushes and such. The fact is, good hands, and bad hands, clump. They have to. If you knew, sure as clockwork, that you would receive AA on every twentieth hand, then so would everyone else. The truth is, you may see AA four times in one session, then not see it for ten sessions. Mathmeticians model this using the Poisson distribution. For example, throw one hundred pennies into the air and let them land on a grid. You will not see one penny per square on the grid, most likely. You will see clumps of pennies and areas of open space. The same is true with hands and roulette (on a balanced wheel).-- Regarding the wheel. There is one mathematical theory that covers it. Bayes theorum. Let's say you divide the board into thirds. We'll call them sections A,B and C. You let the ball roll. No bet. It falls in A. Now, the theorum says that given the larger size of the two areas not hit when combined, it will be more likely that the next roll will hit B or C than A. You then bet those two. The odds will pay you off. You miss, you double your bets. The higher the limit on the table and the lower the initial bet, the more likely that you will come ahead.
Anyway. See ya.

[DPCondit]

[ QUOTE ]
Its my understanding from talking to trainers and stablehands that some predicting factors are not knowable to the general handicapper. That skews results enough so that you can't beat the rake. On the other hand, for the puppies, the weights of the predicting factors tend to distribute themselves in such a way that that the degree of confidence in the prediction can't beat the rake either.


[/ QUOTE ]

In their book, Winning Decisions , Russo and Schoemaker gave horse-race handicappers different amounts of information and asked them to make predictions. First they gave them 5 pieces of information, then 10, then 20, then 40 pieces of information for each horse.

While the degree of confidence in their predictions rose from about 20% confidence steadily up to about 30% with each increasing amount of information, their accuracy did not. Their accuracy peaked out at 20 pieces of information and with 40 pieces actually dropped to its lowest point in the study.

Of course this confidence level only shows the weaknesses of the handicappers, not of the data's predictive ability. The Bayesian Analysis and Neural Net mentioned by Copernicus should not have any such weaknesses. Of course I realize that such study also could not beat the rake either.

Beyond that, I have already stated where I stand on the issue of information not generally available to handicappers, and will not repeat it here.

Good luck,
Don