• Voxitatis Home Page
• About Us • Privacy
Filed at 9:39pm CT, July 17, 2010
Our Delta and Alpha statistics seem to be performing as expected, but Delta is not quite as strong a predictor of a winning team as Alpha is. To test our hypothesis we analyzed every game for which we computed the prediction statistics in the Fall 2009 Season. There were 1,120 games. Then, we looked at how many teams won and lost games for which we predicted a winner at a specified magnitude of Delta or Alpha. Our counts and proportion of a correct prediction are shown in the table below:
As you can see, Alpha outperformed Delta as a predictor of the winner of a given high school game. For example, a total of 140 games over the four weeks had winners with Delta values less than -10, out of 928 games, meaning that 788 games that were predicted to win with Delta values greater than +10 actually did win the game. This gives us a Delta Winning Percentage of 788/928, or about 84.91 percent.
On the other hand, Alpha gave a more powerful predictor of victory at this level. There were 788 winners with Alpha values over +10 and only 93 winners with Alpha values less than -10. This gives an Alpha Winning Percentage of 89.44 percent.
Copyright © 2013. Read our license statement.
By Paul Katula, Voxitatis Research Foundation