Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.1 6/24/83; site dciem.UUCP Path: utzoo!dciem!jeff From: jeff@dciem.UUCP (Jeff Richardson) Newsgroups: net.sport.football Subject: Richardson's NFL Rankings Message-ID: <1158@dciem.UUCP> Date: Tue, 23-Oct-84 12:49:45 EDT Article-I.D.: dciem.1158 Posted: Tue Oct 23 12:49:45 1984 Date-Received: Tue, 23-Oct-84 13:23:16 EDT Organization: D.C.I.E.M., Toronto, Canada Lines: 61 It seems everybody in this news group is writing programs these days to predict NFL games. Well, I am no exception. Actually, my program was written a while ago but I only recently got the data I needed to try it out, and I didn't really write it as a predictor, although it doesn't seem to do a bad job of predicting winners. Anyway, the idea behind this program is that a team's won-lost-tied record doesn't really give you a true indication of how well that team has been doing because it doesn't take into consideration who they've played. I believe I've solved that problem by awarding "good points" for wins and "bad points" for losses. (a tie counts as half a win and half a loss) The number of good or bad points awarded for each game depends on who the opponent was, using the idea that wins against good teams are the most valuable and losses against bad teams are the most detrimental. These good and bad points are then used to calculate a rating that in most cases is roughly equivalent to what the team's won/lost percentage would be if they had played average competition. Also, looking at the good and bad points can tell you a lot about a team's character. A team with a large number of good and bad points (example: St. Louis) has probably been involved in a few upsets, while a team with a small number of good and bad points (like San Diego and New England) has been very consistent. Anyway, without further ado, here are the rankings: Last This Week Week GP W L T PCT GOOD BAD RATING 1 1 MIAMI 8 8 0 0 1.000 8.2 0.0 1.000 2 2 SAN FRANCISCO 8 7 1 0 0.875 5.6 0.9 0.789 4 3 LA RAIDERS 8 7 1 0 0.875 5.2 0.5 0.789 3 4 WASHINGTON 8 5 3 0 0.625 5.7 1.1 0.779 5 5 DENVER 8 7 1 0 0.875 5.1 0.8 0.760 6 6 NEW ENGLAND 8 5 3 0 0.625 4.0 0.4 0.720 9 7 NY JETS 8 6 2 0 0.750 4.4 1.5 0.679 7 8 SEATTLE 8 6 2 0 0.750 3.8 1.0 0.672 15 9 ST. LOUIS 8 5 3 0 0.625 5.3 2.8 0.651 12 10 DALLAS 8 5 3 0 0.625 4.5 2.1 0.646 13 11 CHICAGO 8 5 3 0 0.625 3.8 2.1 0.605 8 12 PITTSBURGH 8 4 4 0 0.500 4.5 3.6 0.555 11 13 KANSAS CITY 8 4 4 0 0.500 3.0 2.2 0.548 10 14 NY GIANTS 8 4 4 0 0.500 3.4 2.8 0.539 14 15 LA RAMS 8 5 3 0 0.625 3.3 3.0 0.519 17 16 PHILADELPHIA 8 4 4 0 0.500 2.7 2.5 0.513 16 17 SAN DIEGO 8 4 4 0 0.500 1.6 2.4 0.448 22 18 INDIANAPOLIS 8 3 5 0 0.375 1.6 2.8 0.429 18 19 NEW ORLEANS 8 3 5 0 0.375 2.0 4.3 0.356 19 20 ATLANTA 8 3 5 0 0.375 1.9 5.3 0.288 23 21 DETROIT 8 3 5 0 0.375 1.5 5.1 0.283 21 22 TAMPA BAY 8 3 5 0 0.375 1.4 5.2 0.269 26 23 CINCINNATI 8 2 6 0 0.250 0.6 4.4 0.263 25 24 GREEN BAY 8 1 7 0 0.125 1.3 5.6 0.236 20 25 CLEVELAND 8 1 7 0 0.125 1.1 5.7 0.221 24 26 MINNESOTA 8 2 6 0 0.250 1.1 6.1 0.200 27 27 BUFFALO 8 0 8 0 0.000 0.0 5.2 0.185 28 28 HOUSTON 8 0 8 0 0.000 0.0 7.3 0.056 Since the program ignores points, it can't pick spreads, but to pick winners, just take the highest rated team of the two. Using that method, it was 10-4 in week 8 and correctly predicted the following upsets: Chicago over Tampa, Detroit over Minnesota, and Rams over Atlanta. If nobody complains, I'll post my rankings every week. (without the long explanation at the beginning) -- Jeff Richardson, DCIEM, Toronto (416) 635-2073 {linus,ihnp4,uw-beaver,floyd}!utcsrgv!dciem!jeff {allegra,ihnp4,linus,decvax}!utzoo!dciem!jeff