Review of the 2012 NFL Season

We've come to the conclusion of the NFL with one of the best Super Bowls I've ever seen. In predicting games against the spread, NFLpickles had yet another great season. That's now 4 winning seasons in the last 5 years! As it has been in other years, our biggest wins this year came after week 12.

2012 NFL Picks Against the Spread Performance
2012 ATS Performance

In other years, including our losing year last year, we've come way ahead in the final weeks and during the playoffs. This year's playoffs picks did 57%, although we're averaging 63% ATS in the last 5 years during the playoffs.

Our ability to work with bankroll to optimize revenue and decrease risk has proven successful. For example, in 2010 although we averaged 52%, we still had a positive 17% ROI since we weighted bets more heavily at the end of the season. Overall, we have a 55% ATS without counting the experimental 2007 year which averaged 61% ATS.

Year W L Pct ROI
2008 22 17 56% 111%
2009 38 23 62% 67%
2010 25 23 52% 17%
2011 23 31 43% -46%
2012 33 22 60% 44%

NFLpickles During the OFF-Season
Every NFL season you learn new things and become aware of opportunities and improvements. During the off-season I will continue to improve the predictive model by testing various approaches to strength of schedule, improving defensive and offensive coefficients, and determine rules to deactivate picks based on injuries, weather, noticeable psychological components, and other things that can significantly change the outcome of the game and is not accounted for in our computerized NFL picks.

A lot of people have told me to build a player-level model and I agree I should. There are various problems with this: 1) it may not guarantee better results, 2) it's hard to get the data, and 3) it will take too much time. So that's why I want to write down rules that will take a game out of contention for picks. This way we have an unbiased approach to determining which games to scratch out. For example, the Green Bay/Minnesota playoff game should have been a scratch because of QB injury.

I will also slice and dice the data during the off-season and provide a few trends that are worth looking at. Many sites have trends that include 6-10 games, that's not a trend. We will be looking at various offensive and defensive components and where opportunities exist that contain a large sample size.

As always, we'll return next year posting unofficial picks from week 1 and then start tabulating results in week 5 of the 2013 NFL season.

Have a great rest of the year everyone. Make sure to check out the site once a month to see what progress we're doing with the NFL against the spread historical stats.

Comments