|
To predict the winners of football games, the opposing teams must be ranked. A number of elements may be taken into account, including yards gained passing and rushing, average yards per down, and quarterback ranking. However, only two things really matter: (1) a team's win/loss record which in turn depends upon (2) the number of points scored by and against that team. A winning team must win games and to do so it must score more points than its opponent. The correlations for 24 seasons between Forecaster's ranking system and the teams' offenses, defenses (determined by points) and win-loss-tie percentages are quite strong.
Using only scores, NFL Forecaster assigns each NFL team an offense and defense ranking and a total ranking. Based on these rankings Forecaster predicts winners of games and the point spreads. Year-By-Year Accuracy for 24 Seasons
Removing the unusual 1987-88 strike season from consideration raises Forecaster's wins and spreads accuracy about 0.3%. Wins/Spreads Accuracy for 24 Seasons
Forecaster's ability to predict winners and spreads is always better than 50/50. There are, of course, seasonal fluctuations. For example, for the 1996-97 season Forecaster predicted winners by 6 points with 100% accuracy but with only 60% accuracy for the 1993-94 season. NFL Forecaster takes into account home-team advantage. However, it gives every home team the same advantage. It does not keep track of a team's win-loss record at home and away and give it a pro-rated advantage (positive or negative) for playing at home. That is, Forecaster uses only scores to rank teams and it ranks them objectively. Therefore, it cannot take into account that the particular team is very good on Monday night or that a quarterback broke his arm during practice, but you can.
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||