NFL franchises use contextualized data to create competitive advantages. In order to realize an edge, teams need to employ the right data in the right way at the right time. This means distilling, interpreting and applying only the most influential data in a framework that accounts for their personnel, their opponents and their evolving game situations. My goal is to be *your analytics department. Each week this season, I want to work for you by giving you a peek into which numbers flag in my models as the most impactful ... or the most misunderstood.*
As always, let me know if your eye test is picking up on something interesting, or if there's a stat/trend you'd like me to take a deeper look at. You can hit me on Twitter @cfrelund. As with any great analytics department, the more collaborative this is, the more value we can create.
Now, let's dig into the relevant data in Week 3 of the 2019 NFL season:
IMPACTFUL DEVELOPMENT: The fallout from a string of QB changes
Just two weeks into the new season, five teams are in flux at the game's most important position:
Jacksonville Jaguars: Starting rookie sixth-rounder Gardner Minshew due to Nick Foles' broken collarbone.
New Orleans Saints: Starting Teddy Bridgewater (or Taysom Hill?) due to Drew Brees' thumb injury.
New York Giants: Starting rookie first-round Daniel Jones with the benching of Eli Manning.
New York Jets: Starting Luke Falk due to Sam Darnold's mononucleosis and Trevor Siemian's season-ending ankle injury.
Pittsburgh Steelers: Starting Mason Rudolph due to Ben Roethlisberger's season-ending elbow injury.
So, how do I code for quarterbacks with little-to-no NFL data -- or very little recent NFL data, in the case of Bridgewater?
One of the things I spend a lot of time putting into my model: past tendencies of defensive play-callers (and systems) against rookie and backup quarterbacks. How likely is Daniel Jones (and Saquon Barkley) to face a five-man front on first down, based on Todd Bowles' past tendencies and the Bucs' current personnel? How often on third down should we expect 49ers defensive coordinator Robert Saleh to show one coverage pre-snap ... only to flip the script in an attempt to trick Mason Rudolph? All of this is pertinent intel.
The idea is to model the defensive minds on the other end, and then find as many comparable situations for the quarterbacks as possible (ideally, circumstances that closely approximate the offensive pieces the QBs currently have surrounding them). Then I do it for each of the rest of the games this season and ... BAM -- I get a ton of shifts for the entire NFL playoff picture. (By the way, I update all of this every week and I'll have to an OK sample size in a few weeks -- likely right around the time when Drew Brees and Nick Foles return to action.)
New Orleans Saints (currently 1-1, in a three-way tie for first place in the NFC South): Before Brees' injury, the Saints were slated to win the division by my model, reaching the playoffs in 62.7 percent of simulations. After Sunday's loss to the Rams, and projecting the Saints will be without Drew Brees until Week 10. they still project to win the South -- but only earn a postseason berth in 53.6 of season sims. Brees is worth 2.1 wins over the next six games, a challenging guantlet that includes road trips to Seattle and Chicago, as well as a home game against the Cowboys. With five divisional games after their Week 9 bye (and the whole division having the bulk of games against each other later in the season), the NFC South is unlikely to be decided until late this season and projects to yield only one playoff team, according to my model. Atlanta is the most likely divisional foe to pass the Saints, but the Falcons would probably have to win at least three games that they are not currently projected to win in order to do so. Matt Ryan and Co. have the best odds of making playoffs if they win three out of the following five games where they have the lowest win probability: at Texans, vs. Rams, vs. Seahawks and at least one of the matchups against the Saints (Brees is projected to be back for both).
Pittsburgh Steelers (0-2, sharing the AFC North basement with Cincinnati): Before Week 2, my model projected the Steelers to win the AFC North, and they reached the playoffs in 54.4 percent of simulations. While an 0-2 start and a QB change have only dropped Pittsburgh's playoff odds to 49.9 percent, that figure now ranks third in the division, behind Baltimore (51.3) and Cleveland (50.1). Now, according to my model, Rudolph provides the least drop-off from the season-opening starter of any of the five teams featured above. I ran the model with Roethlisberger at QB for the rest of the year, just to see what would happen, and the Steelers finished with the second-highest playoff odds in the division: 50.5, right between Baltimore's 51.1 and Cleveland's 49.2. And this reveals that, unlike the NFC South, the AFC North has three teams truly vying for a postseason bid. If the Steelers' defense can stop giving up so many big plays -- and the addition of Minkah Fitzpatrick, which is factored into the above projections, should help on this front -- Rudolph can indeed put them in position to contend for the playoffs.
TWO TRENDS TO TRUST
1) Dak Prescott being a top-eight quarterback. This ranking includes the aforementioned injured QBs and their win-share projections before they got hurt. When I first created my models, I had to establish a time frame for determining when a player has enough of an NFL resume to allow for a true judgment of his skill level. For quarterbacks, my coaching inputs and math inputs were really in lockstep that about three seasons is the magic number. With Dak two games into Year 4, he sits at No. 8 on this season's QB hierarchy. One thing that really stands out is his improvement in the passing game when opposing defenders come within a five-foot halo of his field of vision. Prescott went from earning a first down or touchdown on 23 percent of passes in this situation in 2016 to 48 percent in his past six games.
In the first two Cowboys games of this season, we've seen signs that the fit between Prescott's skill set, Kellen Moore's play-calling and Dallas' current offensive pieces could create even more upside for the 'Boys. The increased use of play-action has helped increase Dak's efficiency. Last season, Prescott only attempted 25.5 percent of his passes off play-action, while this season he's already tracking at 43.5 percent. And according to Pro Football Focus, the Cowboys QB boasts a whopping 157.6 passer rating (158.3 is perfect) off play-action, with 25 completions on 27 throws for 333 yards and four touchdowns.
2) The NFC West starting hot. Before the season began, it seemed like most NFC West narratives had the reigning NFC champion Rams winning a third straight division title, with the Jadeveon Clowney-infused Seahawks representing a potential wild-card team. After two weeks of action, I am not sure where the prevailing media narratives stand, but we're already seeing signs that the entire NFC West is a lot better than advertised.
During the summer, I repeatedly pointed out that NFC West teams projected to produce an abundance of defensive pressures -- one of the most valuable commodities in the game. I also had the Rams and Seahawks as playoff teams, with the Niners right on the bubble (winning nine-plus games in 54 percent of projections). I'm level-setting the past because my preseason projections about defensive pressures/disruptions are tracking favorably, but the model has shifted to the point where it's becoming more likely that the entire NFC West (all four teams) is strong enough to influence the playoffs more than originally projected -- especially since the NFC West faces the AFC North (with no Roethlisberger) and NFC South (with Brees now out of commission for a while).
Here's one note on each team that stands out through the first two weeks of play:
Arizona Cardinals: Despite allowing an NFL-high 22 big plays (10-plus rushing yards or 20-plus receiving), the Cardinals have given up just six more points than they've scored. While the O-line remains an issue, the offense has been able to keep games within reach -- and that's a trend that projects to continue. In this Sunday's Panthers-Cardinals game, my model currently projects only a one-point win for Carolina (24-23; Panthers win in 58.1 percent of simulations). Should Cam be too injured to play, this could easily shift to a Cardinals win due to Arizona's offense. And once Patrick Peterson returns from his six-game suspension, the defense will get a boost.
San Francisco 49ers: Blitz Jimmy Garoppolo at your own risk. The Niners QB currently holds the fifth-best passer rating when facing heat at 139.8, according to Next Gen Stats. Jimmy G has also been lights-out off play-action, as Pro Football Focus has him owning a perfect 158.3 figure, with 15 completions on 19 attempts for 280 yards and three TDs.
Seattle Seahawks: Just like Garoppolo, Russell Wilson is making teams pay for blitzing, completing 20 of his 27 throws for 249 yards and three touchdowns (equaling a 139.3 passer rating, per NGS). The Seahawks are also 4-for-4 on red-zone drives resulting in touchdowns.
TWO TRENDS TO DISTRUST
1) Mitchell Trubisky having zero passing touchdowns. Saying the Bears' offense is struggling is putting it lightly. Chicago currently ranks 30th in scoring (9.5 points per game) and total yardage (263.5 yards per game). So far, we've seen Trubisky complete a lot of quick passes, as 77.1 percent of his completions have occurred on passes thrown quicker than 2.5 seconds after the snap (per Next Gen Stats) for short gains (5.4 yards per attempt, 4.1 air yards per attempt). Part of this is by design, but those numbers are also affected by Trubisky facing major heat in Week 1. In the season-opening loss to Green Bay, Trubisky was sacked five times. But Chicago shored things up before Sunday's win over Denver, as the Broncos failed to record a single sack. And I expect more of this kind of protection going forward. The Bears' O-line -- and how they use blocking tight ends -- should keep Trubisky away from extensive disruption. My model still expects this offense to rank 16th-19th in the metrics most tied to wins, including points per game. And with the team's spectacular defense, that kind of production drives wins.
I went to school, worked and lived in Chicago (still consider it a second home), and I love how passionate Bears fans are. I know some of that passion can trend negative -- especially when it comes to poor quarterback play -- but I'm telling you to stick this one out because it's about to get a whole lot better. And it starts Monday night in Washington. So far, the Redskins' defense has allowed a 64.3 percent third-down conversion rate (the highest mark in the NFL) and six passing touchdowns (second-highest). My model projects the Bears' offense to notch 23 points.
2) The Packers averaging 1.5 points per game after halftime. In last Sunday's win over the Vikings, the Pack scored a TD on each of their first three possessions, racking up 171 yards on just 19 total plays. Then ... Over the course of 11 more possessions, Green Bay didn't score again and managed to net just 164 more yards. The implication here is that their in-game adjustments were ineffective. While it's true that the beginning of games are scripted, this sputtering after the scripted part is far more likely a period of adjustment than a problem that will persist. The Packers have played two divisional opponents that just happen to have top defenses. Meanwhile, Aaron Rodgers didn't play in the preseason, and Matt LaFleur is in his first season as Green Bay's play-caller.
This week, despite facing a Broncos team with former Bears DC Vic Fangio scheming against Rodgers, the Packers project to earn 26 points and 265 net passing yards (they didn't reach 200 in either of their first two games) in part due to an uptick in deep throws. So far, the Packers have only connected on five completions of 20-plus yards. Look for deep passes to be a major factor going forward -- and a big reason why the Packers will score points in the second half of games.
Follow Cynthia Frelund on Twitter @cfrelund.
