Need Help - predictions

[bhatch]

Not sure how to ask, but need a push in the right direction for research.

Is there a mathematical concept that allows you to predict outcomes of matches.

Say you have 4 people named Jim, Scott, Jeff and Kim. In match 1, Jim versus Scott and Jeff versus Kim. If Jim scores better than Scott and Kim scores better than Jeff, and Jim scores better than Kim, you can assume Jim scored better than Jeff. Or in other words A vs B, C vs D and A > B and C > D and A > D then A must be greater than C. So if Jeff and Jim are against each other, and Kim and Scott are opponents, I would like to see if assumptions can be made on who would win in these other matches.

Are there other assumptions that can be made? Is there such a theory or concept? What would it be called so I can start reading about it. Thanks.

[jemidiah]

You're probably looking for some part of game theory (http://en.wikipedia.org/wiki/Game_theory).

Predicting the outcome of matches is difficult. Let's say you're talking about Tennis. If Scott is horrible, Jim will win by a huge margin, probably larger than the margin Kim wins by. Your approach would say that then Jim will beat Kim since his score was higher than hers--but in fact it's more complicated than that, since Scott is horrible. Let's say we account for that by having Kim play Scott--whoever of Kim and Jim beats Scott by the larger margin is the better player, right? Who knows. Again, finding a decent model to predict outcomes of games is difficult.

[opus]

Welcome to the VBForums

You have an error in your first manual try!
On what assumption did you base the following:
Or in other words A vs B, C vs D and A > B and C > D and A > D then A must be greater than C.
C could have scored better then A!
As long as you have some kind of score values for each player, you could put them in a score list (like in any kind of sports Football (American or the real one)......). Using that scorel list, a player on better place in the list will most probably win against a player who is lower on that list.

[namrekka]

Do you mean something like Bayes?

http://en.wikipedia.org/wiki/Bayes'_theorem

[Milk]

A couple more links that might be useful...

http://news.bbc.co.uk/1/hi/programmes/more_or_less/8062277.stm
http://understandinguncertainty.org/