Wednesday, March 11, 2009

A Log5 Look at the Big East Tournament

I was looking around to see if anyone had done the Log5 calculations for the Big East Tournament. Running the seeds out in 3 phases creates something of a challenge for the computation, but ends up with easier-to-understand results. It turns out John Gasaway (aka The Big Ten Wonk) over at the Basketball Prospectus recruited his friend (and former Basketball Prospectus contributor) Ken Pomeroy to run the numbers. I suspect the complications related to the tournament seeding "forced" John to move his Big East Tournament projections off to a blog posting, located under the title "Big East Tournament Preview: A Pomeroy Sighting". Note, the Log5 calculation is an exercise in conditional probability, not a Nostradamus-induced quatrain. It measures probabilities of a set of outcomes relative to each other, and should not, therefore, be taken as an absolute prediction of the future.

The Wonk explores a paradox in the computations in which Louisville has favorable computations in the quarter and semi finals, but Connecticut nevertheless has the strongest rating to win the final. The key, as Gasaway points out, is in the combination of opponents each will encounter on the way.

As proof of fallibility, consider that 2 of the dogs yesterday, DePaul and St. John's won out over their favored opponents (Cincinnati and Georgetown respectively). That popping noise in New York City yesterday was bursting bubbles.

No comments: