Updated 12/4/15 with the lines for the Bengals and Colts games.
Overall Against the Spread: 82-71
Week 2: 9-7
Week 3: 5-11
Week 4: 10-5
Week 5: 5-6 (2 pushes, 1 not bet due to correct line)
Week 6: 8-6
Week 7: 6-7 (1 not bet due to correct line)
Week 8: 6-6 (2 pushes)
Week 9: 6-7
Week 10: 10-4
Week 11: 6-7 (1 push)
Week 12: 11-5
This Week’s Predictions (line in parentheses):
Green Bay Packers at Detroit Lions (+3.0): Predicting 23.5-22.1. Bet on the Detroit Lions.
Houston Texans at Buffalo Bills (-3.0): Predicting 19.8-24.0. Bet on the Buffalo Bills.
San Francisco 49ers at Chicago Bears (-7.5): Predicting 16.5-24.3. Bet on the Chicago Bears.
Cincinnati Bengals at Cleveland Browns (+10.5): Predicting 26.4-19.3. Bet on the Cleveland Browns.
Baltimore Ravens at Miami Dolphins (-4.0): Predicting 23.4-24.6. Bet on the Baltimore Ravens.
Seattle Seahawks at Minnesota Vikings (even): Predicting 19.8-22.3. Bet on the Minnesota Vikings.
Carolina Panthers at New Orleans Saints (+7.5): Predicting 29.8-22.9. Bet on the New Orleans Saints.
New York Jets at New York Giants (+1.0): Predicting 23.3-25.0. Bet on the New York Giants.
Jacksonville Jaguars at Tennessee Titans (-2.5): Predicting 20.8-23.3. Line is correct; do not bet.
Arizona Cardinals at St. Louis Rams (+7.0): Predicting 25.2-20.6. Bet on the St. Louis Rams.
Atlanta Falcons at Tampa Bay Buccaneers (-1.0): Predicting 22.5-23.8. Bet on the Tampa Bay Buccaneers.
Kansas City Chiefs at Oakland Raiders (+2.5): Predicting 25.5-23.2. Bet on the Oakland Raiders.
Denver Broncos at San Diego Chargers (+4.0): Predicting 24.4-21.2. Bet on the San Diego Chargers.
Philadelphia Eagles at New England Patriots (-10.5): Predicting 18.6-30.5. Bet on the New England Patriots.
Indianapolis Colts at Pittsburgh Steelers (-7.5): Predicting 20.0-25.3. Bet on the Indianapolis Colts.
Dallas Cowboys at Washington Redskins (-4.5): Predicting 18.8-23.5. Bet on the Washington Redskins.
Have you done an analysis to see if you could do any better by betting a subset of the games, maybe just those where the model thinks the spread is wrong by a lot? One would think you would do better in those games, but conceivably you could do worse as well if the reason it’s off by a lot is due to injuries?
I’ve looked at it and it’s not real conclusive. Here is a sample:
Over 30+ years of data,
If the spread and my prediction differ by less than 3, I am 2615-2283 (53.4%)
If the spread and my prediction differ by 3-6, I am 1082-1055 (50.6%)
If the spread and my prediction differ by 6-9, I am 272-246 (52.5%)
If the spread and my prediction differ by more than 9, I am 44-32 (57.9%)
It looks like there my be a U-shape going on. My model beats the spread enough to make a profit (52.4% of the time) when the difference is very small (which probably signifies no significant injuries or “news” that my model doesn’t know about). If the difference is moderate (3-9), my model doesn’t make much of a profit, probably because there is an injury or “news” the line is reacting appropriately to (and which my model does not react to, unfortunately). If the difference is large though, I think the line may be OVER-reacting to an injury or something like that, and my model does well again by ignoring the news. Thoughts?
Interesting – would not have guessed that especially after throwing out the (very few) games where the difference was large that your highest return would be from games where the difference was very small.
Another question I thought of – are they getting better at making the spread? Also are you going off the initial spread, or the spread after they’ve changed it to even out the betting?
It’s harder to make a profit in the 2000’s and onward than in the 80’s and 90’s, so yes, I think they are getting better at making the spread. I’m using the closing spread, not the opening spread, because that is the data I have.