Here’s the honest truth preview of Cowboys-49ers
Is Brock Purdy as good as his numbers say he is? Which is the real Dak Prescott; the one from Monday night or the NFL’s interception leader? Almost every conversation about this weekend’s titanic clash revolves around these digestible narratives. Those in the know, know it’s much more nuanced than the simple answer. There are eye-test anecdotes people tend to rely on; but when was their last eye exam, anyway?
No, the numbers tell a much more complete story. Analytics spit in the already-mentioned eye of the common refrain of lies, damn lies and statistics. They can’t explain what’s going to happen every time, but they certainly frame the picture of what is most likely to occur. So for this Cowboys-49ers rematch and latest chapter in an epic rivalry, it only makes sense to dive into the analytics to answer the opening two questions.
For that, we brought in the big gun. Daniel Houston, better known on Twitter by the handle CowboysStats (a must follow), lent us his insight on the fascinating QBs, their offenses and the how they intersect with each other in a detailed Q&A session.
Topic: The 49ers offensive attack and Purdy's place in it
(Photo by Ezra Shaw/Getty Images)
Cowboys Wire: Let’s start with the elephant in the room. Has Kyle Shanahan built the world’s most QB-friendly offense? Do the metrics have insight on how much of their success is independent of Brock Purdy and are there numerical clues as to what works against their attack?
Cowboys Stats: I think he has. Just about every quarterback who’s played in the Shanahan offense — whether a Jimmy G backup, high draft pick or Brock Purdy — has benefited from elite production on yards after the catch, even after accounting for down, distance and target location. This is an area where play design seems to be having a big impact on how this offense moves the chains.
That said, Purdy has so far excelled at completing passes on third down and executing the offense in a way that other Garoppolo backups have not. His completion rates on 3rd down appear to be driving his EPA accuracy component in this case, even as his CPOE is fairly middling. If that continues on Sunday, the Cowboys defense could be in for a tough day.
There are some opportunities for Dallas, though.
Purdy has been fairly sackable, doesn’t pose much of a threat as a scrambler and is most comfortable working the short and intermediate parts of the field. He’s also unlikely to maintain an interception rate this low (2.4%) going forward, given that he does take the occasional risk downfield. A defense that can apply consistent pressure and play soundly in coverage near the sticks has a shot to frustrate him, I think.
Next Topic: Dak Prescott's true self
(AP Photo/Ashley Landis )
So let’s shift the focus to the Dallas scheme and QB combo. You’ve been as vocal as I have about what the real narrative should be this season. Can you bring us up to speed about what Prescott has been?
Cowboys Stats:
Yeah, what bothers me the most about the narrative this year is how Prescott’s entire season has been viewed through the lens of his interception tally alone. The more we hyper-focus on the 15 passes that got picked off, the more we overlook what Prescott has been doing on the other 97% of his plays. And the stuff that hasn’t gotten as much attention has been pretty special.
Despite playing behind a shaky offensive line, Prescott continues to navigate pressure and avoid sacks at the league’s fifth-best rate. Despite largely unreliable options downfield, Prescott has been able to consistently find his receivers in high-value locations.
On third downs, when the defense generally knows he’s going to pass and the entire drive is on the line, Prescott targeted receivers nearly 4 yards beyond the sticks on average. That’s the second-farthest downfield of any quarterback this year. Partly as a result of this knack for exploiting defenses, he has the league’s fourth-best success rate in these drive-saving situations.
At the same time, these risks don’t always pay off. Seven of Prescott’s interceptions this year have come on must-pass third downs. Two more have come on 2nd-and-18+, and another two picks happened in 2 minute drill situations before halftime. That leaves four interceptions on somewhat “normal” first and second downs.
So with some positive results and some negative ones, how do we make sense of Prescott’s risk profile this year? How do we know what he’s doing is effective on balance?
My preferred tool for this is an expected-points model, which considers the down, distance, field position and other information to estimate the impact of each play on a team’s scoring expectancy. And while Prescott’s interceptions have been costly, that impact is dwarfed by that of the other 97% of his plays. Across all situations, Prescott ranks No. 5 overall in expected points added per play, even though the formula heavily penalizes QBs for turning the ball over. By moving the chains and preventing numerous punts, Prescott is providing more than enough value to offset the interceptions.
Supporting Casts
(Photo by Thearon W. Henderson/Getty Images)
Cowboys Wire: So one of the things you mentioned of Prescott’s weapons. When looking at EPA on targets, how far ahead are the 49ers skill position guys compared to Dallas’? How big is the gulf Prescott has to overcome to keep pace with Shanahan’s wizardry?
Cowboys Stats:
The thing that strikes you about the 49ers receivers is how their success is widely distributed.
The best receiver on either team this year has probably been CeeDee Lamb, by a handy margin. He’s earned about 40 more targets this season than the leading San Francisco receiver, and has shown a ton of versatility getting open in valuable spots all over the field.
But beyond Lamb, the 49ers have a clear edge in receiving corps depth, and they seem almost perfectly tailored to make the most of Shanahan’s designs.
Their top wideout, Brandon Aiyuk, can beat you past the sticks and run. Star tight end George Kittle can beat you near the sticks and run. And every time Deebo Samuel and Christian McCaffrey catch the ball 4 to 7 yards short of the line to gain, they’re a threat to pick up the first down — and sometimes a whole lot more.
It’s a tall task for Dallas, which as an offense asks a lot more of its QB. The best way to beat an offense predicated on picking up yards after the catch is to win the completion battle. If the Cowboys defense can pressure and frustrate Purdy into completing fewer passes — particularly on high-leverage third downs — and Prescott continues finding receivers downfield, there’s no reason the Cowboys can’t make up the gap.
Bleeding Edge Insight
Cowboys Wire: I know you tend to scour all of the various analytics takes and folks that are doing some interesting things that might not yet be mainstreamed. Are there any interesting tidbits about these teams you’ve seen that caught your eye, that might become key on Sunday?
Cowboys Stats:
Yeah, there’s a ton of great new work being introduced every year. Timo Riske of Pro Football Focus has done some compelling analysis recently breaking down receiver performance into the main three phases of play: the route, the catch point and the after-catch run. ESPN Analytics released a new receiver metric this year that uses a similar framework with tracking data. This work has informed a lot of what we’ve discussed above.
I’ve also taken a particular interest lately in Twitter user @LuckyProphet5‘s EPA components analysis for QBs. His method basically breaks down the “how” behind EPA, adding context that lets us tell a deeper story than the raw averages do.
When we look at how Purdy has produced his EPA, it becomes clear that there is something bigger going on than simply Shanahan’s offense. His results have also been inflated by a particular context-adjusted completion category which, like EPA itself, is highly responsive to what’s happening on third downs.
Purdy has completed 70.4% of his passes on third down, which is 5.8 percentage points higher than the typical QB would, given the same target locations and situations, according to nflfastR’s completion probability model.
This would be troubling, if not for two things: 1) a QB’s third-down completion rates are volatile and tend to eventually regress closer to his results on first and second downs; and 2) far from overperforming expectation across the board, Purdy has actually underperformed his expected completion rates on first and second downs by 2.5 percentage points.
If Purdy can maintain these abnormally high completion rates on third down, which are some of the toughest situations to pass, the Cowboys might be toast. But instead, I think it’s actually an opportunity for Dallas to dial up pressure, fluster Purdy with some solid coverage looks, and force worse outcomes in a way that can nip a couple extra 49ers scoring drives in the bud.
Tying it All Together
(AP Photo/Michael Ainsworth)
Cowboys Wire: So after crunching all of the numbers of what the puzzle pieces would look like in the majority of scenarios, what’s your overall opinion of what will take place on Sunday?
Cowboys Stats:
For one thing, we shouldn’t expect a repeat of Prescott’s near-perfect wild card round performance. That kind of game is special, and doesn’t come around very often.
That said, I expect Prescott will continue to show full command of the offense at the line of scrimmage and when surveying downfield, even against a very stout 49ers defense.
I also expect the 49ers to move the ball a lot better against the Dallas defense than Tampa Bay did in the previous round. The Cowboys defense has been pretty exploitable throughout the second half of the season, and this 49ers offense has the potential to stretch them to their limit in very different ways than the Bucs were capable of last week.
If the Cowboys defense is back to its old ways from the first 12 weeks, the game should be close. I think people forget that Shanahan, as brilliant as his offense is, has not been able to sustain results like this in the past with a backup QB. The small sample size and third-down dependency of the early Purdy era would make it tough for me to have confidence that he can keep this up indefinitely, if I were a 49ers fan.
But the Dallas pass defense is a big unknown at this point as well. They still face a lot of the same personnel limitations in the secondary that some pretty underwhelming opponents have been exploiting routinely since Week 13. Defensive coordinator Dan Quinn has found better answers for those limitations of late, but we’ll see if it’s enough on the road against a very tough San Francisco.
