/* ]]> */
Mar 192010

Depending on your predictive model, Kansas had somewhere between a 20% and a 25% chance of winning the tournament, and so did Duke.  By winning last night, Kansas’ odds didn’t really change much, since they were supposed to beat Lehigh with near certainty, and the same goes for Duke vs. Ar-PB.

Meanwhile, in whatever pool you’re playing some 40% to 50% of its entries picked Kansas to win the championship, but only 6% picked Duke.  While you didn’t know the exact number prior to yesterday, you had pretty strong clues.  Since Monday night,  ESPN had been showing over 40% of its entries had picked Kansas as the champ, and Yahoo! had nearly half!  In other words, the probability that someone actually picked Kansas was twice as much as the probability that Kansas might actually win.

Let’s say you’re in a 20-entry pool and 10 have picked Kansas, 2 picked Duke.  Ex ante, each Kansas entry has a 2.5% shot at winning, while each Duke pick has a 12.5% shot.  Picking Kansas, in effect, was picking the early round upset lottery:  if you didn’t get many of yesterday’s upsets, you’re already screwed.

 Posted by on March 19, 2010 at 11:19 am