Report Card: NFL Over/Under Win Totals and Fantasy Football
John Ewing, Director of Research and Analytics, takes a look at how well PredictionMachine.com did at fattening your wallet with season win totals as well as giving you an edge in fantasy football.
PredictionMachine.com has offered season win total predictions for each of the last three seasons. In 2010, this analysis accurately projected the top five preseason O/U plays including projecting the New York Jets (9.5) and Kansas City Chiefs (6.5) to go "Over" their win total lines and the Seattle Seahawks (7.5), San Diego Chargers (11.0) and Cleveland Browns (5.5) to go "Under" their win total lines. These top five picks covered by a combined total of eight team wins. In 2011, we had a "2X" pick on the Jacksonville Jaguars "Under" (6.5), which covered easily; while all preseason O/U team win total picks went 18-12-2 (60% O/U).
After the success we had in 2010 and 2011, we had strong reasons to believe that 2012 would be no different. We offered opinions on all 32 teams, NFL O/U Wins Picks, as well as offering 13 normal or better plays (picks that cover greater than 57%). In those 13 “normal” picks we went 7-5-1 (58.3% O/U). The push was on Miami under 7 wins. That play could have been a win for some if you found the line at 7.5 which is what it opened at back in May last year.
Best Calls: Redskins over 6 wins; finished with 10 and hosted a playoff game for the first time in thirteen seasons. Saints under 9.5 wins; finished with 7. The Bountygate scandal clouded the future of a team that went 13-3 the year before. Vegas was not sure exactly how much to move the line with so many uncertainties.
Mulligan: Eagles over 10 wins; went 4-12. The Eagles scored 280 points, the fewest since Andy Reid’s first season. They also committed 37 turnovers and had a -24 turnover differential, tied for last with the Chiefs in the NFL. Injuries played a large part in dooming this pick, left guard Evan Mathis was the only offensive player to start all 16 games.
Final Grade: B+
Normal or better picks (picks that cover greater than 57%) went 7-5-1, 58.3% O/U. All picks combined, because I know some of you are addicts and bet everything, went 15-12-5, 55.6% O/U. Two of the pushes (both under plays on Dolphins and Giants) could have been wins for some as the lines opened half a game more than what we posted. We gave ourselves a B+ because, while it was a profitable year, it was no 2012 MLB O/U wins picks where when we went 9-1, 90% O/U for normal+ plays.
The table below shows all of our over/under season total picks and the results for each.
The best way to grade how well we did at giving fantasy advice is to evaluate ourselves against the two largest fantasy portals, ESPN and Yahoo. We compared our preseason rankings to theirs and looked at who did the best. Rankings using standard scoring were compared as opposed to PPR or auction ranks.
To compare results we looked at preseason rankings, then found the average difference between preseason and final rankings using standard scoring. For example, when looking at quarterbacks ESPN average difference between preseason and final rankings was 5.8 spots with a standard deviation of 4.64 spots. To determine which portal did the best we focus on standard deviation. When comparing the spread of two separate data sets that have approximately the same mean, the data set with the smaller standard deviation has a narrower spread of measurements around the mean and therefore usually has comparatively fewer high or low values. Or as Paul would say, it is more important to get the core right than to hit on some and whiff on others. We did not remove players who were injured because injury risk is a factor to consider when ranking players.
Quarterbacks – Top 20
ESPN: average difference – 5.8, standard deviation – 4.64
Yahoo: average difference – 5.8, standard deviation – 4.84
PredictionMachine.com: average difference – 5.1, standard deviation – 4.00
Our best call: Michael Vick being outside of the top 10, he was #16 in our rankings. ESPN had Vick #7 and Yahoo had him as their #6 QB. Honorable mention to Josh Freeman, we had him at #10 in the preseason (ESPN #16, Yahoo #17) and he would have finished in the top 10 (finished #13) if he didn’t free fall in the final three weeks, 2 TDs and 9 INTs.
Our worst call: Phillip Rivers, #8 in our rankings, finished #21. ESPN ranked Rivers #10 and Yahoo ranked him #9. Rivers passed for 3,606 yards his lowest total since 2008. He also only threw 26 touchdowns, lowest since 2007.
Running Backs – Top 30
ESPN: average difference – 16.4, standard deviation – 20.4
Yahoo: average difference – 16.6, standard deviation – 20.9
PredictionMachine.com: average difference – 16.6, standard deviation – 14.2
Our best call: Maurice Jones-Drew, #9 in our rankings, #5 in ESPN and #6 in Yahoo’s rankings. This was a good call because if you drafted off of our rankings then you avoided taking MJD in the first round, someone else then in your league most likely did. Avoiding a bust is just as good as drafting a stud.
Our worst call: Fred Jackson, #10 in our rankings (ESPN #14 and Yahoo #13), finished #39. Jackson was coming off an injury (fractured fibula) that saw him only play in 10 games during the 2011 season. Prior to the injury though Jackson was having the best season of his career. An injury to a knee ligament in week 1 derailed Jackson’s 2012 season. Even if he had not gotten injured, CJ Spiller may have eroded his fantasy value regardless.
Wide Receivers – Top 30
ESPN: average difference – 17.3, standard deviation – 18.5
Yahoo: average difference – 17.6, standard deviation – 18.4
PredictionMachine.com: average difference – 18.2, standard deviation – 18.6
Winner: ESPN and Yahoo, means and standard deviation are approximately similar. Are you noticing a trend? It is really difficult to predict individual player performance. All three portals were off by at least 17 spots in the rankings for WR.
Our best call: Larry Fitzgerald, #10 in our rankings, finished #41. Wait, what, that’s a big difference. Correct, but it was still a lot closer than ESPN and Yahoo which both ranked Fitzgerald #2. We did not like the quarterback situation in Arizona and while we believe Fitzgerald is still a great WR he cannot overcome the inadequacies of Kevin Kolb, John Skelton, Ryan Lindley, and Brian Hoyer.
Our worst call: Jordy Nelson, #2 in our rankings (ESPN #13 and Yahoo #15), finished #29. Injuries plagued Nelson all season, Jordy only played in 12 games. While we did not expect the touchdown total from last season, 15, to be the same we did expect his receptions and yardage to repeat. If it wasn’t for injuries Nelson very well could have finished in the top 5. He averaged 6.1 targets per game this year (6 tpg last year) and 15.2 yards per reception (18.6 ypr last year).
Tight End– Top 20
ESPN: average difference – 10.4, standard deviation – 7.88
Yahoo: average difference – 10.5, standard deviation – 8.54
PredictionMachine.com: average difference – 10.6, standard deviation – 8.65
Our best call: Greg Olsen, #5 in our rankings (ESPN and Yahoo both ranked him #19), finished as the #6 TE. Finished with career highs in receptions (69) and yards (843), as he became a consistent member of the Panther’s passing game.
Our worst call: Tony Gonzalez, #8 in our rankings (ESPN #7 and Yahoo #9), finished #3. The ageless wonder continues to impress, his 93 catches was good for 2nd among TEs, only 7 WR had more catches then him. His 930 yards receiving was good for only the 5th highest total in his career.
Kicker– Top 10
ESPN: average difference – 9.1, standard deviation – 5.24
Yahoo: average difference – 10.2, standard deviation – 8.33
PredictionMachine.com: average difference – 10, standard deviation – 9.12
Our best call: Matt Bryant, #3 in our rankings (ESPN and Yahoo had him outside the top 10), finished #3. Bryant converted 33 of 38 attempts; he scored double-digit fantasy points in nine games.
Our worst call: Ryan Succop, #6 in our rankings, finished #22. ESPN and Yahoo had him outside of their top 10. He converted 28 of 34 field goal attempts but only had 17 extra point attempts; the Chiefs scored the fewest points in the NFL. Compare that to Matt Bryant who had 44 extra point attempts.
Defense– Top 10
ESPN: average difference – 8.9, standard deviation – 6.56
Yahoo: average difference – 8, standard deviation – 6.02
PredictionMachine.com: average difference – 9.5, standard deviation – 6.47
Our best call: Cincinnati, we had them ranked #7 while ESPN and Yahoo didn’t rank them in their top 10s. Cincinnati finished #4 in fantasy. In our Power Rankings they were #8 in our defensive pass efficiency and #9 in defensive rush efficiency.
Our worst call: Chicago, we had them #10 while both ESPN and Yahoo had them as top 5 defenses. Chicago finished #1 in fantasy. They finished #1 in INTs, #8 in sacks, #2 in forced fumbles, and #1 in defensive touchdowns.
Final Grade: Pass
When you add up all the average differences between preseason and final rankings for each position it looks like this: ESPN average difference 12.6 spots and a standard deviation of 15.1 spots, Yahoo average difference 12.8 spots and a standard deviation of 15.5 spots, and PredictionMachine.com average difference 12.9 sports and standard deviation of 13.8 spots. If you remember from the beginning of this article, when the means are approximately the same then the sample with the smaller standard deviation has a smaller spread of values or in this case spots in the rankings.
You may be wondering why we switched from a letter grading scale to a Pass/Fail system. The average difference is essentially the same for each portal and while we have a better standard deviation, we would expect the performance at each position for each portal to vary from year to year. Therefore I would give each portal a grade of Pass.