To the original poster: Back in the days before the Port Security Act and the fall-out it caused in sending money offshore for gambling purposes, there existed Tradesports.com. It enabled in-game betting and operated as a binary market. For two years, this was my sole source of income and enabled me to pursue a full-time master's degree in OR. As my OR skills developed, a colleague of mine and I built an API interface to the market linking our models to the exchange, automating obvious bets and sounding alarms for judgements requiring human intervention. From my experiences, I think OR is a very powerful tool for live sports betting. Another fun yet short-lived opportunity involved finding arbitrage opportunities between Tradesports and Betfair.com. Both enabled in-game betting, yet one was a binary market and the other decimal. Often, after a score, there existed an arbitrage opportunity as the prices found a new equilibrium. This can probably still be done today if you can find two sites offering in-game betting. Af far as standard non-in-game betting, I found American Football to be the best in terms of modeling success. I have worked on a model for over 4 years and after the first few games of the season, my accuracy is well north of a coin-toss. Anyway, long story short, if you are looking for a partner knowledgable in the application of OR in the world of sports betting (assuming it is legal of course), I am all ears. After a two-year stint in the corporation climate as an OR consultant, I am ready to be indepedent again... Also, I recently discoverd Matchbook.com enables in-game betting.

The best way to make money on sports betting is to act as the bookie. Most betting systems only have a useful positive expectancy over a large number of trials, which is prohibitive using your own money. As a bookie you can take as much risk as you are willing to manage, and can set your own prices.

Of course I wouldn't expect an OR person to go into business themselves as a bookie, but there are many bookies who have very primitive risk management and modelling capability who could benefit hugely and quickly from analytic expertise.

If you go to Google Scholar and search <"horse racing" interfaces> and <"horse racing" "management science">, you'll find some references to work related to wagering systems and whether the efficient market principle in finance applies.

Another point: when you want to make money with betting modelling, predicting the probabilities of the various outcomes is only half the picture - the other half is deciding whether to bet, and how much to bet given your estimates and confidence. Sometimes it actually makes sense to back the team you expect to lose, simply because the odds are higher than they should be.

Wayne Winston is posting World Cup predictions and rankings on his Mathletics blog.

http://waynewinston.com/wordpress/

Studying the stochastic processes of a soccer game is really fascinating. Think of number of different parameters the model would have to incorporate for predicting one game. I would definitely be interested in the science of it.

I highly recommend the book Efficiency of Racetrack Betting Markets. Its a collection of papers on betting markets, and especially horse racing. It has lots of weak stuff, but there are some quite good ones in there. Has quite a few models, and also wagering systems. If you're doing betting modelling its pretty useful.

If you want to get inspired, google "bill benter", he is pretty successful at it (see, e.g., this article from Wired).