NP: shuffle play

The Wall Street Journal takes a look at the NCAA tournament, but I'm not sure if I buy into this whole "outperforming their seeding" notion that forms the basis of the analysis. It seems to assume historical consistencies for some teams, and that simply doesn't make a lot of sense to me.

Anyway, once again I'll be hypothetically trying to win our hypothetical office pool for the tournament using some sort of mathematical model, because actually doing it would be illegal, and that would be wrong. And I may finally have access to some of the stats I need to factor in the things I think are most important, namely rebounding.

This time, I'll try to remember to, you know, write it down. Last year, I didn't, and I ended up winning the whole enchilada, which was all I could really afford given how many people (hypothetically) never paid the guy (hypothetically) running it. I'm pretty sure it used seed, wins and losses -- both the total numbers and the percentage, I think -- and maybe points per game, but I don't remember anything about the actual matchup formula except that I squashed the difference between seedings by something like a factor of 4.

What I've got in mind, if I can get it, is to estimate each team's points in a matchup by taking some number in between one team's points scored and the other team's points allowed. Where each value falls out in between the two will depend on other factors. For example, maybe you get the percentage of the difference based on your winning percentage. We'll see. The possibilities are pretty endless, and they still have to give me a bracket that mostly makes sense. Not too many upsets, but not too few, either.