Crunching The Numbers Returns Week 4

What up BGN, it’s been a while! It’s almost time for football, which means it’s almost time for Crunching The Numbers, my home-brew stat-based ranking system. For those who may be new to the site, this is a side project that I started back in 2011 when I became dissatisfied with “pundits” and “experts” who fell in love with hot teams in September only to see them fall apart in November. I became convinced that there had to be metrics that, if teams performed well in those areas, they would ultimately succeed even if they got off to a slow start. On the other hand, if teams didn’t perform well in those metrics, they would ultimately fail. For the first season I kept my project to myself to see how I did, and in my second I started posting them in the FanPosts. Eventually Jason (the site manager before Brandon) took notice and promoted me to front page contributor. The rest, from there, is history.

But enough rambling on by me. The formula below highlights the findings of my most recent research, which was detailed over two posts earlier this summer. If you didn’t read them, I suggest at least skimming through them first. After four weeks of football I’ll start to post the results on this site since I’ll have a decent sample size. For simplicity, I’ve split up the formula into the offensive and defensive parts depicted below:

OFFENSE

The offensive portion of the formula involves the following six metrics. Rationale on why I picked each one can be found in the links above. The metrics are:

• Yards per pass attempt [YPA]
• Interception percentage [INT%]
• Sack percentage [SCK%]
• Fumbles per game [FM/G]
• Rushing first downs per game [R1D/G]*
• Points forced per game [PF/G]

*In the previous posts I mentioned I would look into passing first downs per game as well to make sure that the positive trend was not a result of the fact that good teams get more first downs overall than bad teams. While there was also a positive trend for passing first downs, it was roughly half as significant, indicating that rushing first downs have a stronger connection to winning.

Those metrics are combined in the following manner:

OFFENSE = (YPA + R1D/G - FM/G)/(SCK% + INT%) + PF/G

For anyone who might feel that adding the percentages is less effective than multiplying them, I do this to avoid a divison-by-zero error in the event that a team avoids a sack or interception over the first quarter of the season. While the error is still possible if a team avoids both a sack and interception, this scenario is far less likely to occur.

DEFENSE

The defensive portion of the formula involves the following six metrics. Rationale on why I picked each one can be found in the links above. In the linked posts, my first attempt to find good defensive metrics had mixed results, but I reiterated my process until I found ones that worked. The metrics are:

• Yards per attempt allowed [YPAA]
• Third down conversion percentage [3D%]
• Passing touchdown percentage [PTD%]
• Rushing percentage [RSH%]
• Takeaways per game [TK/G]
• Points allowed per game [PA/G]

Those metrics were combined in the following manner:

DEFENSE = (PTD% - RSH%) + (TK/G - YPAA*3D%) - PA/G

It should be noted that passing touchdown percentage had a negative correlation to winning, which essentially indicates that the more a team’s opponent scored by passing, the more often that team won. Since it is a percentage, it does not mean teams that allowed more passing touchdowns were more successful.

OVERALL

Finally, the total score is calculated using a highly sophisticated equation that is sure to give your old calculus professor fits:

SCORE = OFFENSE + DEFENSE

And that’s all there really is to it. Stay tuned in to BGN and come Week 4 I’ll debut the first edition of this year’s rankings. But for now, feel free to add your own input in the comments below.