You could take their season's stats (last 52 weeks would be better than just last 2 months) as 50% of their number + last 5 years playing at that course as another 50% of the numerator and then apply your divisor to see what the predictions would be.

I would bet KJ Choi would move further up as an example, because he has done well at Tampa.

I made these factors up on the fly, but you can analyze this to calc a better number.

I think Vegas odds are usually meaner than they should be for most players. I picked Furyk to win the US Open a few years ago when he did, and it only paid out 30 to 1. In hindsight, it was a good bet, but I would bet that people aren't making a living on golf bets, except when Tiger gets going on a roll.

Can you do this for every tourney so I can us it for the golf challenge I participate in? Thanks!

And good for you: starting next week, The Numbers Game will run on Tuesdays, not Thursdays. So it could work. But c'mon, the formula is easy enough. Go ahead and run it yourself, Lazy Bones.

The waxing and waning of a pro's performance week to week seems to me to be statistically confounding. That's one of the reasons I'm not sure season wins is an appropriate component of any predictive formula.

Had I the brains or time to try to come up with a predictor I would probably want to factor past performance at the specific golf course in question along with performance in the last one or two events.

Of course, were any method possible, it would have been discovered by now and nobody in Vegas would be making book on golf tournaments!

You know, if you could factor in the course attributes, like length, fairway width, green size, whether the rough is penal or not, etc, that might be interesting to look at. I'll bet it would affect your predictions.

What would really be cool is if you could apply the course attributes to an entire season of stats, to see how much a particular course-type affected various types of players. If that worked out, it could lead to a tool that course designers might use.

Of course, that's assuming you're remotely correct in how you've chosen to apply the formula. Dave's 40-30-20-10 rule was calculated after an event to show the relative weight of a few common statistics (GIR, PA, DD, DA) on a week-to-week basis. Dave then applied the rule to entire seasons and got even better results (for obvious reasons).

You've applied it backwards. While Dave used it as an analytical tool (after the fact), you've tried to cram it into the mold of a predictive tool. If golfers were consistent, that might actually work. They're not.