Brock Purdy Is Only *Sort Of* a "System Quarterback"
Kyle Shanahan and a strong supporting cast can help a QB... but only so much.
The first full season of the Brock Purdy Saga will come to a (merciful?) end at the Super Bowl on Sunday, its narrative either resolving with Purdy a champion — presumably recording an impromptu commercial about going to Disney World — or a late-drafted fraud finally unmasked once and for all by the Chiefs’ dominant defense. (In which case, Taylor Swift will do the Disney commercial instead.)
Truth is, no matter what happens on Sunday, it won’t change the minds of the most ardent Purdy lovers or haters. A lot of folks are already dug in on their preferred narrative about the 49ers’ quarterback: Either Purdy is the most efficient passer in football (I mean, he does have the stats) or, at best, a useful backpack carried around by Kyle Shanahan, Christian McCaffrey, Deebo Samuel and George Kittle.
The latter camp assumes Purdy would be a mediocre QB without his coach and supporting cast. But few have bothered to actually quantify how much Purdy might be benefiting from those factors. How much, for instance, is playing QB for Shanahan actually worth? What do great teammates add to a QB’s output? This post is intended to at least wade in the direction of an estimate, if not provide a final answer.
For the supporting cast question, I called upon my previous research about, ironically enough, Patrick Mahomes and the talent he inherited before his first season as Kansas City’s starter in 2018. There, I measured the quality of a QB’s skill position teammates by adding up a weighted sum of the Approximate Value (AV) his leading running backs and receivers (both WRs and TEs) had produced over the previous three seasons. By that measure, Purdy — playing alongside McCaffrey, Brandon Aiyuk, Kittle, Samuel and Jauan Jennings — had the second-most talented group of skill teammates among all NFL starting QBs (minimum 6 starts) this season; only Philadelphia’s Jalen Hurts worked with better talent.
Score one for the “Purdy is propped up by great teammates” argument. Except, how much do those great teammates really matter to a quarterback’s bottom line? We can estimate this by looking at how a QB’s Expected Points Added (EPA) improves or declines with changes to his supporting cast quality between seasons. Since measuring by AV is somewhat circular for QBs who stick with the same skill group across multiple years — good quarterbacking can create a larger value pie to be split by the offense, and vice-versa — I honed in on the 72 cases since 2000 of a QB starting at least 6 games for a different team in consecutive years.
Among those cases, sometimes an improved supporting cast was associated with a big improvement in EPA/game. But nearly as often, it wasn’t. As it turns out, the relationship between changes in established supporting-cast talent and QB performance is relatively weak:
Put another way, the predictive relationship between playing with great players and putting up monster numbers — which underlies much of the Purdy-skeptical discourse — is only true to an extent. Based on the data above, we’d predict that Purdy would drop from 5.51 EPA/game (No. 5 among starters this season) to 5.25 EPA/game (still No. 5, slightly ahead of Mahomes) if his actual supporting cast was replaced with a league-average one. (He’d dip to 5.13, landing in sixth place ahead of Tua Tagovailoa, if he had to play with Mahomes’ now below-average supporting cast from 2023.)
That’s hardly the drop to the bottom of the league you might expect from a lot of the “system QB” chatter that follows Purdy around. And a similar — if somewhat larger — effect emerges when we try to unpack the Kyle Shanahan Effect on quarterbacks.
Shanahan has been either QB coach, offensive coordinator or head coach (and de facto OC) for nine different QBs at the NFL level — Matt Schaub, Donovan McNabb, Rex Grossman, Robert Griffin III, Brian Hoyer, Matt Ryan, Nick Mullens, Jimmy Garoppolo and Brock Purdy. Of those, all but Mullens and Purdy have had seasons as starters (again, minimum 6 starts) with different play-callers as well.
As we can see from the chart above, Shanahan definitely boosts his QBs’ stats. If we weight by how many seasons our group of QBs had both with and without his coaching using the harmonic mean, that group averaged 3.86 EPA/game in seasons under Shanahan and only 3.15 EPA/game in seasons without him. That difference, 0.71 EPA/game, would be enough to drop Purdy to 4.79 EPA/game (and No. 11 among starters) if we applied it to his 2023 stats.
Of course, it’s easy enough to also see in the chart that Purdy has averaged the most EPA/game under Shanahan of any starter; only Ryan (who won an MVP working with the coach in 2016) is especially close. If Purdy is a product of Shanahan’s system, he has also done the best job executing that system out of any NFL QB who’s tried it.
So what do we get if we apply both the supporting-cast adjustment and the Shanahan adjustment to Purdy’s 2023 numbers? (Assuming both are independent effects, which is probably not the case in reality; after all, playing for Shanahan has boosted the stats of Deebo, Kittle and company for years as well. But we’ll go with it for now.) Subtracting both factors, Purdy ends up at 4.54 EPA/game, which would sandwich him between Kyler Murray and Justin Herbert for 13th in the league, ahead of Trevor Lawrence, Matthew Stafford and Derek Carr.
In other words, even the most severe possible downgrading of Purdy’s season, adjusting away all the measurable benefits of his coaching and teammate talent, still leaves us with a solid, above-average quarterback. And it makes sense: There’s no way an outright bad (or even subpar) passer leads the league in QBR without ability to go with a favorable situation.
Purdy is good — say it, embrace it. Just be ready for plenty of over-the-top reactions to his performance on Sunday, however this chapter of his unlikely story ends.
Filed under: NFL, Super Bowl