bruhgraw@uok.UUCP (11/02/84)
I have written a VERY simple program (in Pascal) to make predictions of NFL games. Basically all it does is read three variables for each team from a database. These variables correspond to (1) the team's record so far, (2) a figure generated by adding its prowess in offense and defense, and (3) a "bias variable" which I use to keep track of such things as winning (or losing) streaks, major injuries (like Kellen Winslow), and the like. The program compares the variables of the two teams playing each other, and whichever one has the higher number of points is the predicted winner, with the spread equalling the difference of the points for the teams. This is probably the simplest application of a prediction algorithm that you can think of, but it's been doing pretty well...it averaged 8.7 games correct for the first half of the season. The program's main feature is that it formats output into a nice, easily- read table, with the picks, spreads, and rankings of each team. It also chooses the game it chooses to be the best one being played that week. Here is the output from it this week. Predictions for week 10 in the NFL: Visiting team Home team Prediction by spread ------------------------------------------------------------------------- ( 1) Miami at (10) N.Y. Jets Dolphins by 20 (25) Cincinnati at ( 2) San Francisco 49ers by 23 (12) New England at ( 3) Denver Broncos by 4 (14) L.A. Rams at ( 4) St. Louis Cardinals by 13 (11) Kansas City at ( 5) Seattle Seahawks by 9 ( 7) L.A. Raiders at ( 6) Chicago Bears by 2 (18) Atlanta at ( 8) Washington Redskins by 11 (13) N.Y. Giants at ( 9) Dallas Cowboys by 11 (15) San Diego at (17) Indianapolis Chargers by 7 (16) Philadelphia at (24) Detroit Eagles by 1 (19) Tampa Bay at (23) Minnesota Vikings by 4 (20) Green Bay at (21) New Orleans Saints by 6 (27) Houston at (22) Pittsburgh Steelers by 20 (26) Cleveland at (28) Buffalo Browns by 11 Game to watch: ( 7) L.A. Raiders at ( 6) Chicago