Stats are like podcasts: their numbers are seemingly endless, and it’s hard to know which one to listen to at a given time. That’s why when Sports Info Solutions came out with its Total Points stat, which had Aaron Rodgers as having created the most points in the NFL this season, eye brows raised.
Who were these people? Why should we listen to them? And what is this statistic truly telling us if it’s accomplishing its stated goal? Come to think of it, what is that goal?
To understand, we have to back up. Sports Info Solutions used to be Baseball Info Solutions before expanding into football four years ago. They were an advanced metrics pioneer, leading the way for stats like Defensive Runs Saved in baseball to become part of our mainstream discussion of the sport.
This was, in some ways, the genesis of Total Points, according to SIS director of football & research Matt Manocherian. How do we parse credit on a given play and assign a tangible value to it? The answer is relatively simple, even if the process isn’t. Alex Vigderman, senior research analyst at SIS, spearheaded the creation of Total Points and you can read his introduction with Football Outsiders here.
The concept starts with EPA, a metric they believe to be cleaner and more all-encompassing of a given play’s value than anything else out there. This may come as a surprise relative to Rodgers, considering EPA and its ESPN cousin QBR don’t reflect the same kind of success this season for the Green Bay quarterback.
But this is where division of credit comes in.
The premise functions much like EPA, where a play’s productivity is measured relative to how much a play like that in that situation would normally be worth. A 2nd-and-6 run for three yards gets a value based on what a three-yard run in that situation is historically worth in generating points.
Total Points attempts to follow the same logic on a more granular level.
“What makes us different from PFF, is we’re not actually grading anything on every play, but what we are doing is consider any sort of event that can happen. So, we can account for a pass that is catchable but not on target, but caught or dropped, on and on and on,” explains Manchocherian, a former NFL scout for the Saints and Browns.
“We’re trying to account for all the different coaching points that coaches really care about.”
They’ve taken it a step further and created a system where they model the expected EPA of a given play and first ascertain the value the actual play created. A slant to the left side has a history of productivity. A lead play to the B gap has an average value, so based on that, how much additional, or how much less value did the play create?
From there, missed assignments or productivity beyond expectation can all be quantified. If Rodgers throws a bubble screen on target, he’ll get a little credit, but not as much as the receiver making a play after the catch. While his yards per attempt will reflect a poor play if Geronimo Allison or Marquez Valdes-Scantling only gets three yards on the screen, it doesn’t hurt Rodgers in Total Points the same way it would in nearly every other metric.
On the other hand, if he fires a post to Davante Adams where he can catch and run, part of the credit will go to Rodgers. The delineation of credit, good and bad, comes from this historical charting data. How much does a blown block on an inside run generally hurt a play? OK, how much did it hurt here? That delta value gets used on that specific play.
In the Adams example, how much would an off-target but catchable ball hurt the expected points of the play? Rodgers would get dinged for that even if Adams made the catch. On the flip side, if Rodgers puts a ball on target and Adams drops it, Total Points will still give him credit for “creating” the EPA of that play because he did his part in the creation of it.
According to SIS, no quarterback lost more value this season to drops than Rodgers, as we discussed a few weeks ago. As Manocherian points out, the dings on Rodgers this season come in drips and drabs, an off-target throw here, or a throw-away there. (He was second in the NFL in throw-aways in 2019, a year after leading the league in them by a mile). But Total Points accounts for interceptable passes, not just interceptions, a place where Rodgers has always scored extremely well.
It’s also worth noting Rodgers would be behind quarterbacks like Lamar Jackson and Drew Brees in a rate stat version of Total Points. On a per-play basis, he wasn’t as efficient as they were, but he played more than Brees and threw more than Jackson, which pads his total beyond theirs. Of course, being on the field matters in the case of Brees, and Rodgers’ health was a question coming in. Playing all 16 games relatively healthy can’t be ignored.
The next step for SIS, which works with a group of NFL teams to provide information, is to push Total Points forward into a predictive metric. Based on what happened on this play, how can a team expect a player to perform on it moving forward? This would be particularly useful, for example, with Rodgers right now as we head into January. Based on this season, what plays would be best to call in the playoffs with the quarterback concerned about the timing of the offense?
Manocherian admits even the resident Packers fan in the SIS offices looks at the numbers somewhat incredulously, but they believe their process is sound, or at least better than anything anyone else has tried. The goal is an admirable one: assign value to the success or failure of a player independent of the play’s outcome. In other words, what was the value of a player doing his job or not doing his job on a given play, relative to the expected value of that play to begin with?
Total Points will take time to become mainstream, if it ever happens. Even EPA, on which Total Points is based, lags behind other advanced metrics in its adoption. But it’s worth noting Total Points also loves other analytic darling quarterbacks from this season like Dak Prescott, Patrick Mahomes, Kirk Cousins, and Lamar Jackson. If it’s seeing what other advanced numbers are seeing with Rodgers as the outlier, we must at least consider that the added context — parsing the credit — reflects something other metrics are missing.