Let’s take a look back at the eight tournaments for which log5 calculations were posted in this space and see how those predictions turned out. What follows is the winner of each tournament, their chances of winning it and how that ranked in the field.
76 Classic: Saint Louis 35.7% (1)
Old Spice Classic: Dayton 19.5% (2)
Battle 4 Atlantis: Harvard 14.6% (3)
Great Alaska Shootout: Murray State 31.3% (1)
Maui Invitational: Duke 43.1% (1)
Paradise Jam: Marquette 56.9% (1)
Charleston Classic: Northwestern 38.6% (1)
Puerto Rico Tip-Off: Alabama 20.3% (3)
The results are difficult to interpret when it comes to validating the predictions. If every top pick had won that actually wouldn’t validate the model very well at all because each top pick had a significant chance of not winning its tournament.
Based on the probabilities given by the log5 analysis, one would have expected just three out of the eight top picks to win a tournament, instead of the five shown. That would suggest probabilities were too low for the top picks, but there is far too little data here to draw any meaningful conclusions. You could find enough things in Duke’s or Murray State’s wins to account for the good fortune of top picks in this batch of tourneys.
The easy winner for least likely occurrence was Norfolk State’s appearance in the Paradise Jam championship game, which was given a 1.5% chance of happening ahead of time. From what I’m seeing, no other possibilities given less than a 10% chance actually happened in these tournaments. It looks like UCF’s appearance in the Battle 4 Atlantis title game, at 12.5% is a distant second in this category.