UMass87 wrote:Actually, Ken Pomeroy has UMass winning 8 of the final 13. He has not been a very good predictor of UMass this year. In fact, his predictions have been so bad that his predictions are probably negatively correlated with UMass' results.

Pomeroy is not one of the better raters. Sagarin and Greenfield are very good. Massey is pretty good. Some others like Self and Colley are also pretty good.

Greenfield and Sagarin provide both predictive and "wins" ratings which turns out to be very informative. Some also provide measurements of standard deviation or stability which I would guess would be pretty high or low, depending on how you look at it, for UMass. For example, Pomeroy's rating of UMass' average performance might be fairly accurate. The average, however, does not give a good picture of a team that is winning games they should lose and losing games they should win. They fluctuate from playing great to poor but the final power rating only gives a measure of average performance. One number can't possibly give a complete picture for something like sports where there is some randomness to the results. For that you need at least two numbers reflecting both the mean and the standard deviation (instability/uncertainty). From that point predictions are statistical which is to say that Team A will beat Team B within some confidence interval. At this point, the margin of error for predicting the result of any UMass game is probably pretty high.

On the difference between "wins" ratings and "predictive" ratings, the former is good at measuring how well a team will do in a close game against similarly rated opponents and the latter is good at measuring a team's overall power and can be affected by how much a team plays up or down to much higher or lower rated teams. BC is an interesting case at this point. Their "wins" rating (See Sagarin's Elo-Chess) is quite high while their predictive rating is not that great. What this says is that BC tends to play down to lesser opponents and would be at risk for an upset in the early rounds of the NCAAs, for example. However, if they survive these potential scares, they might just be good enough to go all the way to the final four. In other words, if they survive week #1 they will likely survive week #2 as well.