It's almost time for football, which means in a few short weeks I'll be debuting the 2015 edition of my "Crunching The Numbers" power ranking system. For those of you who may be new to the site, Crunching The Numbers is a sort of pet project of mine where I attempt to objectively rank each NFL team without taking into account their wins. This has been an ongoing thing for me since the 2011 season, and was partly inspired by the 2009 Denver Broncos, who started the season 6-0 before finishing 8-8. That epic collapse convinced me that there was something that went beyond wins, something hidden deep within the statistics that was a more accurate indicator of a team's likelihood to succeed. I've had mixed but encouraging results since I started.
Every year I try to improve the formula by analyzing the success of the previous year's iteration. The analysis that led to this formula, which can be found on this page, was a resounding success. A number of you provided welcomed and candid feedback about what I could do to improve my analysis, which I will be incorporating into the analysis next season. So while I concede that the efforts which went into this formula were not perfect, they are what I have to work with and I think can lead to a successful season for Crunching The Numbers.
But what exactly am I trying to accomplish with all this? To recycle an old cliche, I am trying to use numbers to separate "contenders" from "pretenders." My benchmark is to be more accurate at predicting a team's playoff success by Week 4 than their score differential at that time. So, without further ado, here's the 2015 mathematical formula that will attempt at accomplishing this feat:
Offensive success is focused around the quarterback's efficiency and putting him in a position to succeed. This mostly entails protecting him in the pocket and having a good running game to convert first downs and run out the clock once the team is in the lead. The offensive half of this metric is divided into three segments.
Passing: Passer Rating/(15.83)
By dividing the passer rating by 15.83, the quarterback will achieve a score between one and ten (since the maximum passer rating is 158.3, and 158.3/15.83=10), which will keep this metric in line with the other statistics used in the formula.
Offensive Line: Rushing First Downs/Game - Sacks/Game
Pretty self-explanatory. Sacks/Game are subtracted instead of divided because if the sacks per game metric is too small, it will artificially inflate this portion of the formula and skew the overall results for that team. By subtracting, the potential range for the offensive line metric is kept in check.
Rushing: Rushing Attempts/Game*Rushing %
Running the ball in today's NFL is simply a tool to run out the clock after the passing game has generated a large lead. Therefore, teams that win tend to run the ball more than teams who are trying to overcome a score deficit quickly. By multiplying by rushing percentage (expressed as a decimal), the rushing attempts will be corrected for teams that run more plays than others.
OFFENSE OVERALL: Passing + Offensive Line + Rushing
Unlike offense, my analysis earlier this summer suggested that an effective defense will force the offense to abandon their game plan, rather than focus on certain areas like the pass rush or run defense. Because of this, winning teams have defenses that affect the playcalling of their opponents. The defensive side was split into two metrics:
Passing: (Passer Rating Allowed/15.83)/Passing % - Sacks/Game
Again, passer rating is scaled down to a range between one and ten and corrected for how often a team passes during the game (expressed as a decimal). Sacks per game are subtracted from this, even though they didn't have a strong correlation with winning last season. But I'm calling that a fluke for right now and will monitor that moving forward.
Rushing: Rushing %
As mentioned above, defense is dictated by playcalling metrics, so the most relevant statistic in this case is the rushing percentage of the opponent (expressed as a decimal). A lower rushing percentage is preferred because it indicates the opponent is passing more to play catch-up.
DEFENSE OVERALL: Passing - Rushing
You may have noticed that the defensive formula is built to make the overall number smaller. That is because it will be subtracted from the offensive number:
OVERALL = Offense - Defense + Score Differential +Turnover Margin/Game
Score differential is the strongest indicator of a team's success and cannot be ignored. Turnovers often either make or break teams (see: 2014 Eagles) and are tacked onto the end to reflect that importance. ...And that's it! These are the numbers I will be using to rank each team this season. I look forward to seeing how the formula performs in 2015 and I hope you enjoy following along on my endless quest to discover the best predictors of team success. I wait until Week 4 to debut the first edition in order to get a decent sample size of statistics, so keep an eye out for them!