Computer simultated games

[ncsucharlie]

Hi guys, I'm writing a java program that simulates games over and over and comes up with a run total average. Has anyone done this before?

I was wondering if there was a way I could pull up to date season stats on the players from my java program, but I can't find any sites that offer that kind of ****. It is a pain in the ass now to enter the team stats for each player.

I just started it and right now it is only based off of the batters at bats, hits, doubles, triples, and homers. Still got to integrate the pitchers into the program somehow (very important)

Here is an example of one game that I simulated 500,000 times, here is the avg scores for Houston/Seattle tonight (plz don't tail this!):

Mariners: 2.9849
Houston: 3.3924

Any java programmers out there that are interested and want to lend a hand or offer some advice let me know, I'll share the source code!

I'm gonna go drink now and see if the Astro's can score exactly 3.3924

[rjp]

I hope you're using some advanced simulation techniques (like those presented in Amazon.com: Monte Carlo Statistical Methods (Springer Texts in Statistics): Books: Christian P. Robert,George Casella).

[ncsucharlie]

wooo! you should have tailed my computer generated pick!

so I guess nobody has done this type of handicapping .. oh well

[rjp]

Originally Posted by ncsucharlie

so I guess nobody has done this type of handicapping

Don't be so naive.

[ncsucharlie]

dude chill out, I'm talking about if anyone on this message board has done anything like this.

from the lack of responses I guess not or nobody cares to contribute

[homedawg]

ncchuck, dont take the comments the wrong way. I know RJP,gets into the mathematical equations of the possible outcome for any event.

I too was on hou yesterday and did not place my bet based on any simmulation.


Keep that system going, and hopefully you can win some $$$$ for all of us...

[flmmkrz]

I know there was a guy on covers who used to post his computer plays. I don't know if he's still doing baseball but he was solid in basketball. You might wanna look him up if you're looking for someone to work with

[rjp]

Sorry about how that post came off.

The questions that come to mind are: what are you looking for out of the simulation as output?

Obviously you want "winners", so to speak, but you're looking to simulate as best possible the approximate odds for the game.

I'm currently working on bayesian models that include simulation to fit the model, but it's not at a point worth betting with.

Also, you'll need some sort of measure for error so that you're not betting on games with a confidence level that covers a large band of probabilities. In the case of MLB games, the probability changes for each game, so you can't exactly use a binomial distribution to obtain a 99% confidence level (something I personally use for specific things so that I don't overbet).

Anyway, I like to see the discussion, so any details about the exact method(s) you're using would be helpful.

[JohnnyMapleLeaf]

DAMN! I should have tailed your computer generated pick!



Katidy does something like that as well I think....