Mathematical Model of the NFL (21 October 2007): » Off Topic » Off Topic » Mathematical Model of the NFL (21 October 2007):
10/20/07 2:45:03PM
Hello everybody.

Last week, my NFL picks versus the spread earned a return on invested capital of +60.6%, and, believe it or not, my compounded rate of return (when all picks are included, with all profits and losses fully invested (equally across all picks) in each new set of picks) is now at an absolutely staggering +133.9% -- that is my ACTUAL (!) compounded Return on Invested Capital since I've started to post the picks of this mathematical model of football.

For those of you who are interested, my new numbers, picks, and predictions for the week's upcoming NFL games are now up... I'm posting them at:

My posts on the numbers underlying the NFL are getting to be longer and longer, and since I didn't think MMAPlayground would feel cool about such long posts being put-up here on his forum, I went-off and started my own blog, where I can talk about value investing -- as it applies to sports.

For those of you who haven't yet seen what I'm doing, I'm attemting to prove that wagering on sports IS a form of value investing, if it is actually approached AS investing. Wagering on sports is really about "finding value" in the offered wagers. Efficiency mis-matches on the field need to be discovered, first, and these, then, in turn, need to be matched-up with inefficiencies in the point spread(s) offered. Each week, there are only a small handful of wagers that are actually mis-priced (undervalued), and these few cases are what I search for. The math underlying this is my own, and I'm not yet fully ready to reveal the underlying concepts. I do believe that I'm the only one attempting this with football -- I've found 2 particularly valueable stats (of my own creation) which seem to have enormous predictive value in determining the outcome of football games, particularly when combined with the idea of adding points given by the spread.

So, if this is the sort of thing that interests you, have a look.

It is long conceptual stretch to consider this kind of thing to be "investing," but rest assured, I KNOW that it is investing, and this is essentially my attempt to prove it.

10/20/07 3:28:32PM
I had some sucess doing this with baseball (of course my model was a bit simpler and I didn't create it I just used an existing one), but the return is much smaller per game. The fact that there are so meny games makes up for that though. I am interested to know, what is your return rate on college vs. pro games? I would think college would be much smaller seeing as there stats are influenced my so meny more factors. Also, have you automated this or are you running the numbers in an excel sheet? Really, what I want to know is what are you using? Maple? some other high level math software? or just a calculator and some scrap paper? I haven't had time to read your whole blog (I will make some time soon), but I might be able to help you improve your output. Not that you need to improve 134% return, but maybe you can make slight adjustment to output more data that will be useful to more people. i.e. fantasy player stat projections. I have alway dreamed of sitting down and creating a more for stat projections and spread betting, but you have it half done. The last thing I want to know is have you taken historical NFL data and tested performance over a larger sample. Something like going back to 1995 and taking data from like the first 5 weeks of the season and predicting week 6 and so on. The only problem would be finding the point spread.
10/20/07 3:41:49PM
No, I haven't been able to back-test it as much as I'd like, because I don't have point-spread data going back very far. Is there any place on the web where point spread info is available for years gone by...?
10/20/07 6:50:43PM
there are odd back to 03 on
They might be a little hard to scrape, but I will keep looking for more.

10/21/07 9:55:55AM
mathematical model of the nfl? how about a mathematical mode of the mma?
Related Topics