All Forums | Systems & Strategies

NCAA Spread Mathematical and Statistical System

...

Participants:

Search

Posts: 41

lokg2getevn

Prospect

Joined: Jan, 2009

Posts: 8

Posted: Jan. 21, 2009 - 12:23 AM ET #26

Freaky, thanks for posting your picks. It takes time to develop and test these systems. I appreciate your ideas and work.

0

FreakyFresh

Prospect

Joined: Dec, 2008

Posts: 498

Posted: Jan. 21, 2009 - 9:46 AM ET #27

Okay guys I want to know what you think of this tweak to the system and get your feedback or input:

1. Well first off I'm sure you guys know the Law of Averages and Law of Large Numbers. After an infinite amount of trials the average of of trials should equal the expected value. Nothing new there, not going to mess with the ancient mathematicans.

2. If you guys know those laws, something similiar is used on the Stock Market to project the rise and fall of different stocks. Many different average prices are tested and compared over different lengths of time to each other. Say that the average of the last five days of a closing price of a stock was much higher than the average closing price over say 2 1/2 weeks. It would be expected then in the very near future for the stock to decline. Vice versa for a shorter average underperforming against a longer times averagem it would be expected to rise in the near future. This may seem confusing but I hope it is clear.

3. I want to note that is one of the very best indicators for a a stock flucuating, even though that there are billions more. Therefore what if I were to take this approach to my prediction indicators?

Say for example I have prediction indicator X. X has predicted games within a standard deviaton of 11.5 points over the course of the past 25 days. Yet, in the past 5 days X has predicted games within a standard deviation of say 10.2. Therefore outperforming it's own self lately. Therefore I would fade the plays slightly for X in the coming days because it would be expected that the following days should miss by a large margin to rise the average back up to 11.5.

I think this can be the art form of my system. I would have to decide how much I want to fade them or tail them but I can mess with it.

Know that was long but any thoughts?

0

dabar

Prospect

Joined: Sep, 2008

Posts: 32

Posted: Jan. 21, 2009 - 12:11 PM ET #28

Stock market price predictor strategems have been tried before for gambling and this is the results I have observed.

1- Re: Law of Large Numbers. Let's refer to flipping a ' fair coin' 100, 1000, and 10,000 times. Theory proposes that the average should near 50 %. In using this idea in sports there occur some problems because the situation does not replicate the theoretical model .

A-There is no 'fair coin' because conditions change. The changes may be easy to observe, and results predictable, ( top player is suspended team winning strea is over) and some not so easy ( starters are not motivated ). With too many condition changes you do not have the stability needed to use the model.

B-The number of games where the conditions are continuous ( always the same) does not approach an infinite amount. A month long stable period is considered very good.Therefore Law of Large Numbers fails to apply and with it Law of Averages.

The results more closely follow a binomial distribution model. There are a few good write ups about binomial distribution elsewhere and would be beneficial to anyone to read them.

2-Your predictor may find a standard deviation of points scored but there is merit to finding a deviation from the set line. For whatever reason the line moves, but if you use that as your norm you do not have to have any inputs on what made the line move- it just is - And this is the critical information that you want as your output, this is the point at which you make your wager.

3- Modes play a bigger part here than on stocks. Say a team's average is larger or smaller than its mode by a clear margin. That would result in a set of predictions that would be different from the averages model.

4-Some gamblers like to chase, therefore I'm always on the lookout for someone who can create a program with perfect Standard Deviation. Perfect S.D. to me would be a program that picks Wake to win today, fails to pick accurately the next, then is accurate on the third, fails on the fourth and is a winner on the fifth and so on. The program would not pick any long winning streaks nor would it be better than 50 % but chasing would be supremely profitable. I'm looking for any insight on this type of predictability pattern such as you would find in a stock that trades steadily between $25 and $30 Any thoughts? Or is this predictable S. D. program unattainable?