I've now tried three weeks worth of predictions. I'm using current performance in 2009 as the basis for these predictions. Thus, there was no prediction for Week 1. In Week 2, I went 9-7 (56%). In Week 3, the record was 11-5 (69%). This past weekend (Week 4), I went 10-4 (71%). So, for the season to-date, my predictions have gone 30-16 (65%). This is higher than I was expecting -- I was actually figuring at ending up around 50%.
My predictions were right for: the Giants, the Bengals, the Redskins, the Jaguars, the Colts, the Bears, the Saints, the 49ers, the Broncos, and the Vikings. I missed on the Ravens, the Raiders, the Bills, and the Chargers.
After the jump, at the request of lebowskibronco, I will share the formula I've been using, and let you all critique it, should you so desire (just try to not laugh too much at it :D)
The method I decided to use shakes out as follows (and fyi, I have this all in an Excel spreadsheet, rather than calculating it by hand):
Points are the ultimate determinant for who wins and who loses. So each team is assigned a "base point" score. These scores are based on the Points For/Points Against of each of the teams in the contest. That score is determined as follows:
((points for / games played) + (opponent's points against / games played)) / 2
Example: In Week 5, Denver will host New England.
Denver's Points For = 79
New England's Points Against = 71
((79 / 4) + (71 / 4)) / 2 = 19 (NOTE: the spreadsheet is set to round calculations)
Denver's Base Points = 19
New England's Points For = 87
Denver's Points Against = 26
((87 / 4) + (26 / 4)) / 2 = 14
New England's Base Points = 14
Base Points are adjusted by each team's win/loss record.
Base Points + (Wins - Losses)
Denver = 19 + (4-0) = 23
New England = 14 + (3-1) = 16
The Record Adjustment score is modified by each team's home/away record, as appropriate.
Denver's at home and their home record is 2-0. Thus, their adjustment is 23 + (2-0) = 25
New England is the visitor and their road record is 0-1. Thus the adjustment is 16 + (0-1) = 15.
4)Division Record Adjustment
If the teams are in the same division, their divisional record is added as an adjustment. New England and Denver are in different divisions, so zeros would be entered into this category & no adjustment made.
Denver's score 25, New England's 15
5)Conference Record Adjustment
If the teams are in the same conference, their score is adjusted by their conference record. If they are in different conferences, zeros are entered in this category.
Denver = 25 + (3-0) = 28
New England = 15 + (2-1) = 16
6)Non-conference Record Adjustment
f the teams are in different conferences, their score is adjusted by their record against the opposing conference. If they are in the same conference, zeros are entered in this category.
Denver = 28
New England = 16
Each team's score is adjusted by the number of games in a row won or lost. If they are on a winning streak of 1 or more, a positive number is added. If they are on a losing streak of 1 or more, a negative number is added.
Denver 28 + 4 = 32
New England = 16 + 2 = 18
Thus, in this example, Denver would be predicted -- based on season-to-date performance -- to defeat New England.
1)It usually takes me between 15 and 30 minutes to enter the numbers into the spreadsheet to predict the games being played on a given weekend.
2)I will be interested to see if the current pattern of slowly increasing accuracy continues as the season progresses.
About mid-week, I'll crunch the numbers for Week 5 & see how that plays out.