Read this for a description of APR
Over the course of the regular season, APR picked 240 NFL games week 2 to week 17. I am counting the Bengals @ Eagles tie game as a push in the following tallies.
Visitor vs Home:
Visitor Won | Home Won | Total | |
Picked Right | 69 | 90 | 159 |
Picked Wrong | 34 | 46 | 80 |
Total | 103 | 136 | 239 |
APR picked the home team to win 124 times (90 right + 34 wrong). Since 136 home teams won, that's pretty good evidence that some kind of factor should be added to favor home teams in generating picks.
Picks by Matchup Ratio
The matchup ratio (ρ) for each game is the quotient of the corresponding power indexes for each team (larger divided by the smaller). It is intended to be an indication of "pick confidence": ρ values near 1.0 involve teams judged to be very close in power, while large values indicate more of a mismatch.
Range | Picked Right | # Games | % | PV+HW | PV+VW | ||
ρ | <1.01 | 5 | 9 | 55.6% | 2 | 3 | |
1.01<= | ρ | <1.05 | 26 | 41 | 63.4% | 9 | 8 |
1.05<= | ρ | <1.10 | 29 | 46 | 63.0% | 11 | 15 |
1.10<= | ρ | <1.15 | 21 | 41 | 51.2% | 9 | 10 |
1.15<= | ρ | <1.20 | 24 | 32 | 75.0% | 5 | 10 |
1.20<= | ρ | <1.30 | 21 | 28 | 75.0% | 5 | 8 |
1.30<= | ρ | <1.50 | 24 | 29 | 82.8% | 3 | 11 |
1.50<= | ρ | 9 | 13 | 69.2% | 2 | 4 |
"PV+HW" stands for "Picked Visitor and Home Won", and indicates the number of games in each range where APR incorrectly picked the visitor to win (and could be picked up by giving home teams more power).
"PV+VW" stands for "Picked Visitor and Visitor Won", and indicates the number of games in each range where APR correctly picked the visitor to win (and could be lost by giving home teams more power).
This indicates that simply giving home teams more power would cost more correctly picked games (where the visiting team wins) than it would gain.
The drop off of accuracy for ρ >= 1.50 is due to the fact that very large values of ρ are only seen at the very beginning of the season, when APR only has a few games to produce power indexes.