The NFL's 100th season is now firmly a part of my historical reference model, which means it's time to go back and analyze each player's individual contribution metric. This is a numerical value I have created that adds (or subtracts) each player's impact on his team's ability to win games, for every snap. The goal is to better understand player value by capturing production in context, such as down and distance, score and time, type of play called and opponent faced.
Undertaking a big contribution-metric deep dive of several historical campaigns -- I look at 10 seasons, six seasons, three seasons and one season in a weighted algorithm -- is also how I focus on building team and player projection models for the upcoming season this early in the year. Finding and applying the most predictive past trends and trajectories for players, position groups and sides of the ball in systems helps shape opportunities and expectations for the upcoming season.
Before we turn the page from the 2019 campaign, though, here is my all-star roster consisting of the top player -- according to contribution metric -- at each position on offense and defense:
The NFL's reigning MVP was my favorite kind of analytics success story -- it resulted from the blend of Jackson's exceptional skill set with the right play-calling and a supporting cast tailored to make the most of his attributes. One example of this is that Jackson threw for 11 touchdowns compared with zero interceptions on play-action passes in 2019 (the best such TD-to-INT ratio in the NFL, per Next Gen Stats). Further, he had 24 touchdown passes (against two interceptions) vs. the blitz, 10 more such TD tosses than any other quarterback, per NGS.
There is a lot of value in earning a new first down on first down, and no player earned more such rushing first downs in 2019 than Henry, who had 31. He also added nine rushing touchdowns on first down (second-most in the NFL) and one more via the pass. Pro Football Focus measured his after-contact yards per rush at 4.2 (the best in the NFL among those with a minimum of 100 rushes, including playoffs), which helps characterize just how much value Henry added to the Titans' offense. And that's before factoring in the impact he had on the rest of the offense via the threat of his production.
McCaffrey led the NFL in scrimmage yards with 2,392 (the next-closest player was Ezekiel Elliott with 1,777) due in part to opportunity and in part to capability. NGS tracks McCaffrey as playing 93.4 percent of the team's offensive snaps (highest in the NFL among running backs). On first down specifically, his 1,262 scrimmage yards were the most for any player by almost 200 (in second was Derrick Henry at 1,098), and he chipped in nine total first-down touchdowns. First-down production and middle-of-the-field work are strongly correlated with drive success (that is, compiling first downs and touchdowns). NGS shows that McCaffrey averaged 5.0 yards per rush inside the tackles (second in the NFL among those with a minimum of 75 inside runs).
My models love Thomas' game, both when he's the subject of a target and as a decoy when he is not. Thomas, of course, set an NFL record with 149 receptions, 91 of which netted his team a first down (most in the NFL), along with nine touchdowns. NGS shows that 97 of his receptions came on short passes, which was 35 more than anyone else. But don't overlook his impact on deeper passes. Thomas caught 68.3 percent of his targets of 10-plus air yards, the best rate in the NFL among those with a minimum of 25 such targets, per NGS. When not the subject of the pass, his ability to attract the opponent's best and/or multiple defensive backs ranked third last season, per my computer vision (as measured by which defenders are near him and their proximity to him).
Texans QB Deshaun Watson is going to miss my model's highest-rated on- and off-ball receiver from 2019. Hopkins, who was traded from Houston to Arizona this offseason, ranked third in receiving first downs with 68 (only Michael Thomas and Julio Jones had more). Interestingly, almost half of those came on second down (30), and 19 of those were earned on second-and-7-plus yards (the most in the NFL).
NGS shows that Kittle led the NFL in receiving yards per route run in 2019 with 3.3 (among those who ran a minimum of 100 routes). This makes sense, given his production as a passing target: Kittle recorded 75.2 receiving yards per game over 14 games, the second-highest rate among qualified tight ends, and the fourth-best catch percentage (79.4%). What takes a little more context to understand is that these great passing stats also accompanied elite impact as a blocker in the rushing game, seeing as how tight ends are usually better at one or the other. My computer vision shows that the average yards per rush gained on the ground when Kittle was in the path of the rusher (that path being defined as the actual route that the rusher ran, with 3 feet added on each side of his route) was the second-highest among tight ends at 5.3 yards per rush.
The tricky part about measuring production at any position on the O-line is that the sum influences each individual component. Sometimes this works in favor of my win-contribution metric, as one player really stands out, with his production shifting the output of the entire unit upward significantly. PFF rates Tunsil as the third-highest tackle in pass blocking (among those with a minimum of 300 snaps), having only allowed 29 pressures in 656 pass-pro snaps (including playoffs). My computer vision metrics add that his individual impact in pass and run snaps combined for the highest win-contribution metric among left tackles by a slim margin. This also increased the Texans' overall O-line production according to my computer vision from right near the bottom in 2018 to 19th overall last season.
Quenton Nelson did not allow a sack on 562 pass-blocking snaps last season, per PFF, helping drive his value as my model's highest-rated left guard. Per my computer vision, Nelson's production in the run game was the second-highest in the NFL among all guards (and the best among left guards) as measured by the number of rushing yards earned when he was in the path of the rusher (4.9).
Kelce made this list a year ago, and now he's here again, with nearly the same results ... In 2018, he kept opposing defenders at least 5 feet from the quarterback on 24.1 percent of passing downs. In 2019, it was 23.9 percent. Both marks were the best at the position. When I use computer vision to measure how players along the O-line provide clean pockets and prevent pressure, it's clear Kelce set the standard for elite the past two seasons.
Opposite what happened with Laremy Tunsil, who didn't have a lot of help, Brooks stood out even though the rest of the unit was also strong. PFF attributes Brooks with allowing only one sack on 605 pass-blocking snaps, and he was their highest-graded guard in 2019 (among those with a minimum of 300 snaps).
Neither Ramczyk nor left tackle Terron Armstead allowed a single snap in the regular season and Ramczyk was the highest-graded tackle overall (all per PFF, among those with a minimum of 300 snaps). My win-contribution metric also ranked Ramczyk as the top tackle.
Since 2016, Aaron Donald has racked up 267 pressures (the most in the NFL per NGS), with 69 of those coming in 2019 (second in the NFL). Donald's ability to get within 5 feet of an opposing quarterback is, annually, where he excels. In 2019, he was double- or triple-teamed more than in any other year in his career, per my computer vision, and managed to still top all interior defenders with 84 trips inside the 5-foot halo of opposing quarterbacks. NGS also adds that Donald unsurprisingly got off the line of scrimmage in an average of 0.88 seconds last season, which was the fastest among interior defenders (of those with a minimum of 200 pass rushes).
Campbell's listed with his current team above (Baltimore), but in 2019, he was one of the few bright spots on the 6-10 Jags. PFF rated Campbell as its third-best interior defender in 2019 (among those with a minimum of 300 snaps), as measured by their grading system. My win-contribution model rated him second-best. The balance in his game -- he ranked third in yards-per-rush allowed in his space (3 feet on either side, all along a rusher's path), and he entered the opposing quarterback's 5-foot halo at the second-highest rate -- created the net impact that Campbell was able to limit opposing offenses at the second-most-efficient rate.
No one had more quarterback pressures in 2019 than Smith's 71, per Next Gen Stats. NGS also shows that he was able to pressure opposing QBs at the second-highest overall rate in the NFL (14.8%) among those with a minimum of 250 pass rushes. Smith lined up on the edge and in the interior, and this versatility boosted his win contribution overall.
PFF ranked the younger Watt as its best edge defender in 2019 (among those with a minimum of 300 snaps). NGS adds that he was able to sack opposing quarterbacks on 3.2 percent of pass rushes, ranking fifth among edge defenders with a minimum of 300 pass rushes.
PFF ranked Davis as 2019's second-best linebacker, but my model has him just sneaking up into the first-place slot, with his versatility leading to stopped opponent drives. With 13 trips inside an opposing quarterback's 5-foot halo (four leading to sacks) and the fifth-fewest yards-per-rush allowed among linebackers when in a runner's path, Davis was linked with the most stopped overall drives and touchdowns among linebackers in 2019.
Kendricks was PFF's highest-ranked LB in 2019 (among those with a minimum of 300 snaps). Coverage is where Kendricks' efforts produced the most success, with just an 83.9 passer rating allowed when in coverage, which is about 20 points lower than the league average against linebackers, per PFF.
Everyone knows to try and avoid Gilmore's coverage unless you have a really good plan ... so while you'd expect him to see fewer targets than most, you'd also expect a relatively high passer rating allowed, since opponents would limit themselves to higher-probability attempts. But this is not the case. PFF counts a passer rating allowed in coverage in 2019 of 47.4, the fourth-lowest in the NFL among those with a minimum of 50 targets. My model adds factors like who Gilmore was covering, along with the play call and situation, which netted out both an association with an increased time to throw for opposing quarterbacks (theoretically allowing more time for a pressure to happen up front for the defense) and an increased level of difficulty in terms of player and play call assigned. These factors drove Gilmore's win-contribution metric up to No. 1 among cornerbacks.
PFF tracked 83 targets thrown to a receiver in his coverage without him allowing a touchdown, which was the most in the NFL. He also recorded a 43.0 passer rating allowed in coverage, which was second-lowest in the NFL (among those with a minimum of 50 targets). The gap between Gilmore and White was the closest for any first- and second-ranked player at a position.
After his Week 3 trade from Miami to Pittsburgh, Fitzpatrick allowed only a 27.3 passer rating in coverage, per PFF, which is a lower number than if a quarterback throws the ball away on purpose (39.6). NGS adds that Fitzpatrick aligned at deep safety on 90 percent of snaps with the Steelers, as opposed to only 22.4 percent in his entire Dolphins tenure.
Adams is listed as a strong safety, so he counts here, but his alignment and positional versatility (three safety spots, corner and pass rusher) provide far greater value than he would at just one position. His 90 pass-rushing snaps (second-most at the safety position, with third being 72, per PFF) netted 23 pressures (most among safeties, per PFF). NGS adds that his 17 quarterback pressures in 2019 were the most among defensive backs, as well.
NGS shows that Mathieu played 98 percent of defensive snaps for the Chiefs, including the playoffs (1,205 of 1,230). Always a playmaker in pass defense, Mathieu's run-stopping ability ranked seventh-best among safeties in 2019, per my computer vision. And the fact that that figure was even better in previous seasons -- when he was a Cardinal and Texan -- reflects a different strategy in K.C., not an erosion in capability.
Follow Cynthia Frelund on Twitter @cfrelund.