Written by: Eric Meyer

Published: December 12, 2018

Over the last few days, I've been perturbed by the use of statistics to stick up for under-performing quarterbacks.  Shannon Sharpe had the gall to suggest Aaron Rodgers has had a better season than Tom Brady.  Other pundits have propped up the statistics of Cam Newton, Eli Manning, and Kurt Cousins to suggest that these quarterbacks are secretly playing well and that our eyes are deceiving us as these players rack up loss after loss.  This has irked me for the better part of a week, so I'm going to write some of my thoughts on this subject.  I'll cover a few statistics that ought to be permanently discarded from meaningful football discussions and replaced with superior metrics.

Completion Percentage

Why it's bad: completion percentage doesn't tell you anything useful.  A completion could be a gain of 5, 10, 20, or 40 yards.  It could even be a loss of 5 or 10 yards.  Why would such a limited stat with no context be used to evaluate quarterback performance?  It should be obvious on its face why this statistic is virtually worthless, and yet it's parroted repeatedly.

What you should use instead: Success rate

A play is deemed successful if it helps pick up first downs.  A successful play picks up 40% of yards needed on first down, 67% on second down, and 100% on 3rd/4th down.  This is clearly a better way to measure efficiency because it includes the most crucial element of football: the ability to move the chains.  Completion percentage treats a 2 yard pass on 3rd and 12 the same as an 8 yard pass on 2nd and 6.  Success rate gives a much clearer understanding of how often a quarterback makes positive plays.

Passer Rating

Why it's bad: Passer rating was revolutionary in its time.  However, this statistic has a ton of problems.  For one thing, the parameters are arbitrary.  The value for each element appears to be at the whim of the creator with no statistical explanation or validity.  Secondly, it ignores sacks, which kill drives and result in high turnover rates compared to any other play type.  Thirdly, as mentioned earlier, it has no control for context.  An 8 yard completion on 3rd and 15 is graded as a better play than a 4 yard completion on 3rd and 3.  In short, this statistic is fatally flawed and should be replaced, but with what?

What you should use instead: DVOA

DVOA is a proprietary statistic of FootballOutsiders.com.  It stands for "Defense-adjusted Value over Average."  DVOA contextualizes every pass play, determines how successful it was, and compares it to the average quarterback performance in similar situations.  It relies heavily on success rate while giving additional credit for big plays and larger penalties for disastrous plays like interceptions and fumbles.  Context matters in football, and it should matter in football statistics as well.  DVOA reflects that, while passer rating does not.

TD:INT Ratio

Why it's bad: This is probably the most lazy attempt at a "catch-all" stat for quarterbacks, but I hear it uttered so frequently that I'm lucky I haven't pulled all of my hair out.  A good quarterback likely drops back to pass around 600 times per season and we choose to evaluate their performance based on 6-8% of their throws?  TD:INT ratio is a tidbit.  It's a nice little piece of information.  It has no value in evaluating quarterbacks because it ignores over 90% of the plays a quarterback makes (or doesn't make) over the course of the season.

What to use instead: Anything else

Seriously, it's hard to imagine a more worthless rate statistic than TD:INT ratio.  If you want some examples of better numbers, look at ANY/A, Expected Points Added, or even ESPN's mysterious QBR.  Stay away from TD:INT ratio.  It's bad for you.

My final thoughts on this are simple.  Trust what you see.  If your team can't score points, it's okay to question the quarterback even if he's completing 70% of his passes and has a 90+ passer rating.  If your team is averaging 5 yards per carry on run plays and you're still struggling to sustain drives, your quarterback is probably not playing as well as the box score indicates.  More advanced metrics help us see when a quarterback is actually performing and when he's just stat padding.