January 2, 2024
Here is a preliminary calendar of games for January 2024 representing roughly the first 325 games of the NCAA D3 Men’s Volleyball season. We have noted some games that you will not want to miss in black (Premier) and blue (Prime). We have also included preseason ratings on the calendar. The Premier and Prime classifications are based on the sum of the ratings of both teams.
I am curious. Do the ratings in the above table consider for Home vs. Away? ( I am guessing not) I suspect the home team warrants something in the interval (0,.5] while the away team might be in the interval [-.5,0). Certainly not the same for every team in reality – very cumbersome dealing with that! I typically create P(Win) functions using a logistic function of the differences in ratings; k=-.3 for P(Set Win) & k=-.6 for P(Match Win), which gives roughly the same result as a binomial interpretation of the P(Set Win). i.e. First to win 3 of 5. What function do you use if forecasting the chances for a match win based on your ratings? (If logistic, my guess is k<-.6 by a little bit since your ratings are more variable than mine to start the season?)
Accounting for no home/away adjustments, we are on opposite sides of P(Match Win) = 50% in just 18 of 289 D3 matches. (~6%) Not too bad, really. Of course, the ratings will change from day to day as matches get played, so this may not continue. I suspect as the year goes on this will dip below 5% of the roughly 1400 matches played, and even if one model is a better predictor of those outcomes, the sample size will likely be too small to guarantee significant differences, I think.
Love your site! Keep it up.