What I learned while doing 2019 fantasy football projections

Troy Wayrynen/USA TODAY Sports

If you're not already aware, I'm the guy who does the player projections here at ESPN.

Although some would swear that the projection numbers are totally automated or randomly generated, I assure you they are not. I have a lengthy process that involves both statistical calculations and subjective inputs. The latter is where this piece truly comes in handy. To begin each league year, I go team by team and thoroughly analyze league, team, coach and player trends. From there, on the player level, I generate projected dropback, carry and target shares for each player.

I recently completed that process and -- same as in recent years -- I took notes. Below are my observations, as well as a short application to fantasy football in 2019.

1. Over the past 10 seasons, only four offenses have averaged at least 3.75 offensive touchdowns per game in a season: the 2018 Chiefs (4.11), 2013 Broncos (4.05), 2011 Saints (4.00) and 2011 Packers (3.82). Interestingly, the Broncos (3.29), Saints (3.31) and Packers (3.06) all averaged more than 3.0 scores per game the following season. As we head into 2019, the question is not if the Kansas City offense will fall off but rather the extent to which it will. History suggests that the Patrick Mahomes- and Andy Reid-led offense will score roughly one fewer touchdown per game but will still rank near the top of the league.

2. Expanding on that a bit, 36 offenses averaged at least 3.0 touchdowns per game during the nine seasons spanning 2009-17. The average output the next season for those 36 units was 2.7, which is still well above the 2.30 league average in the span. Only six of the 36 offenses were below that average, and 13 repeated at or above the 3.0 mark. In 2018, the Chiefs (4.1), Saints (3.5), Steelers (3.2), Rams (3.2), Colts (3.1), Seahawks (3.1) and Patriots (3.1) each cleared 3.0 offensive scores per game, and history suggests we can again count on good-to-great production from these teams.

3. On the other side of the coin, we have the offenses that have struggled. During the same 2009-17 window, 21 offenses were at or below 1.5 touchdowns per game. The 21 units collectively averaged 1.3 scores per game, but that number jumped to 2.1 per game the following season. Seven of the offenses were above the aforementioned 2.30 league average, and only two repeated at or below 1.50 per game (2013 Jaguars, 2011 Chiefs). This supplies some optimism that the 2018 Cardinals (1.5 per game) and Jaguars (1.4) will be quite a bit better in 2019.

4. Cardinals RB David Johnson is averaging a 75 percent carry share and 17 percent target share in 40 games since taking on lead back duties in Week 13 of the 2015 season. Arizona has a new coach in Kliff Kingsbury, and Johnson disappointed in 2018, but the 27-year-old is ticketed for a workhorse role and is thus a solid bounce-back candidate.

5. Running back units under Falcons head coach Dan Quinn haven't eclipsed a 17 percent target share since 2016, and backs under new Falcons offensive coordinator Dirk Koetter haven't eclipsed a 16 percent share since 2015. Devonta Freeman's 10 percent target share back in 2017 (his last full season) is what we should anticipate in 2019, with Ito Smith near the 6 percent mark he soaked up in 2018. Both could be limited a bit in PPR, with Atlanta focused on getting the ball to its strong wide receiver trio.

6. Ravens QB Lamar Jackson set a single-season quarterback record with 147 carries last season. Jackson was the starter for only the team's final eight games and averaged a massive 16.0 carries per game. The question is, is that workload sustainable long-term? Consider prolific rushing quarterbacks Michael Vick and Cam Newton, who have played 13 and eight NFL seasons, respectively. Vick played one full 16-game season and never cleared 123 carries in a single campaign. Newton has been more durable, appearing in five 16-game seasons (though he's two for his past five) and never clearing 139 carries in a season. Jackson could certainly change the game and roll past 200 carries, but that's a risky bet. This is also worth keeping in mind for Josh Allen's prospects, as he was on a 130-carry 16-game pace last season.

7. Speaking of the Lamar Jackson-led Ravens' offense, consider the positional target shares in Jackson's eight starts: 56 percent wide receiver, 30 percent tight end, 14 percent running back. Both the WR and RB numbers are well below league average, and the TE rate is 50 percent above the league rate. Second-year TEs Mark Andrews and Hayden Hurst figure to benefit.

8. The Bills' offense has leaned heavily on the run in each of Sean McDermott's two seasons as head coach, and each of offensive coordinator Brian Daboll's past five offenses have been run-first (if not run-heavy). Especially with inaccurate Josh Allen under center, expect the Bills to lean heavily on the legs of Allen, LeSean McCoy and newcomer Frank Gore this season.

9. Bengals RB Joe Mixon appeared in 14 games last season and handled a massive 78.5 percent of the team's designed runs in those games. If his 16.9-carry and 3.9-target-per-game role carries over to 2019, Mixon will have a hard time not posting a top-10 fantasy campaign.

10. Cowboys WR Amari Cooper handled target shares of 22 percent, 22 percent and 19 percent when active during his three full seasons in Oakland. Last season, he sat at 13 percent prior to a Week 8 trade to Dallas. Not only did Cooper average a career-high 25 percent share in 11 games with Dallas, but he also never fell below a 16 percent share in a game and was fantasy's No. 9 wide receiver from Week 9 on.

11. Lions WR Marvin Jones Jr. has averaged a target share of at least 19 percent each of the past four seasons (three with Detroit). Jones missed just under half of last season with an injury, but the 29-year-old is ticketed for a full-time role opposite Kenny Golladay again this season. He's in the WR3 mix.

12. Packers QB Aaron Rodgers has missed at least seven games in two of the past six regular seasons. Of course, he appeared in all 16 games in the other four and has played in 15-plus games in nine of the past 11 seasons. Rodgers is 35, but durability is an overrated concern, and he's coming off a season in which he threw only two interceptions on 597 attempts. In fact, Rodgers is the only player in NFL history to throw fewer than four interceptions on 450-plus attempts in a season.

13. There were six games last season in which Randall Cobb was out and rookies Marquez Valdes-Scantling and Equanimeous St. Brown were both active and operating as Green Bay's No. 2 and 3 wideouts behind Davante Adams. MVS was on the field for 231 of 261 pass plays and handled a 17 percent target share in those games. St. Brown was in for 167 of 261 pass plays and handled a 10 percent share. A lot can change in an offseason, and the duo will need to compete with the likes of Geronimo Allison, Jake Kumerow and J'Mon Moore, but early indications are that Valdes-Scantling has the leg up for No. 2 duties among the second-year receivers.

14. If Texans RB Lamar Miller doesn't handle at least half of Houston's designed runs in 2019, it will be the first time he fails to eclipse 50 percent in the category since his rookie season in 2012. Miller was at 55 percent (and 10 percent of the targets) during 15 games last season, which allowed him only one top-10 fantasy week. Expect to see more D'Onta Foreman this season.

15. Texans WR Keke Coutee played at least two-thirds of the snaps in five games during his rookie season. His target totals in those games were 15, seven, five, nine and 14. That's an average of 10.0 and works out to a team share of 28 percent. Either Will Fuller V (three) or Demaryius Thomas (one) were active behind DeAndre Hopkins in four of those five games, so it wasn't simply a product of being forced into heavy usage. Coutee is an obvious breakout candidate, especially if the Houston offense figures out how to turn field goals into touchdowns.

16. Frank Reich of the Colts has been a head coach or offensive coordinator for five seasons. Each of those five offenses distributed a target share of at least 23 percent to the team's tight ends. That includes a share of at least 27 percent each of the past three seasons. Granted, Reich has benefited from top-end talent at the position (e.g. Antonio Gates and Zach Ertz), but he maximized Eric Ebron in 2018. The Colts will be even stronger at the position in 2019 with a healthy Jack Doyle in the fold. Note that Doyle ran 151 routes and was targeted 33 times, compared to 101 routes and 22 targets for Ebron, in the six games both were active last season.

17. Jaguars WR Dede Westbrook handled a 22 percent target share as a rookie and a 21 percent share in 2018. Overall, he's averaging 6.4 targets per game. Jacksonville's abysmal quarterback situation held Westbrook to one top-10 fantasy week last season, but Nick Foles offers an upgrade at the spot for 2019, and Westbrook still has little competition for targets. He's well positioned for similar volume and improved efficiency.

18. Over the past 12 regular seasons, Andy Reid-led running back units are averaging 17.3 offensive touchdowns per season. That includes at least 11 rushing scores five of the past seasons and at least three receiving scores all but one of the past 12 seasons. Although he certainly benefited from home run talent via Brian Westbrook, LeSean McCoy, Jamaal Charles and Kareem Hunt during the span, Reid has tended to feature a lead back, as opposed to utilizing a committee. It's probable that Damien Williams, Carlos Hyde or a rookie will emerge as the team's feature back in 2019, but it's tough to determine who it will be. All will be high-risk, high-reward fantasy picks.

19. Chiefs WR Sammy Watkins averaged a 21 percent target share (6.9 per game) during the 10 games he played in full last season. In fact, Tyreek Hill's 73 targets were only four more than what Watkins saw in those weeks. Eight of the 10 games came during the regular season, and Watkins was fantasy's No. 16 scoring wideout those weeks. Watkins' durability is an obvious concern -- he has missed at least three games in three of the past four seasons -- but he plays a big role in arguably the league's best offense, which puts him in the WR2 discussion.

20. Back in 2017, Keenan Allen played 84 percent of the snaps, ran a route on 91 percent of the team's pass plays and handled a 28 percent target share. Excluding a Week 15 game last season, in which he was injured early, Allen was on the field for 85 percent of the snaps and 91 percent of the pass plays and handled a 28 percent target share. How about that for consistency? Once riddled with various injuries, Allen has appeared in all 34 of the Chargers' games the past two seasons and is about as reliable as they come at the position.

21. Rams RB Todd Gurley II was injured in Week 15 last season. At that point, he had played 86 percent of the snaps, handled 74 percent of the designed runs, run a route on 69 percent of the team's pass plays and racked up a 16 percent target share. That's compared to 82, 71, 71 and 18 percent, respectively, during 16 games in the 2017 season. Put another way, Gurley was playing almost an identical role to 2017 prior to an injury that derailed the rest of his season. Only 24, Gurley might come at a discount in drafts this year, but he still has elite upside.

22. Perhaps the most overlooked aspect of Sean McVay's offense is the enormous number of created scoring opportunities. The Rams' offense scored 2.71 touchdowns per game in 2017 (fifth-most) and 3.16 in 2018 (fourth-most). That in itself is impressive, but consider that the unit averaged 2.65 (second-most) and 2.68 (most) field goal attempts per game, respectively. In fact, those numbers both rank top-five in the NFL over the past decade. Combining the touchdowns and FGAs into one number ("scoring opportunities") results in 5.35 in 2017 (second in the NFL) and 5.84 in 2018 (best in the NFL -- yes, higher than the Chiefs' 5.7 mark). In fact, Los Angeles' 5.84 in 2018 ties the 2016 Falcons for the second-best mark of the past decade behind only that of the 2011 Saints. Incredibly, McVay's 2016 Redskins offense posted a 5.3 mark in the category, which was fourth in the league and the 20th-best mark of the past decade. If you're not buying the McVay hype, you're missing something.

23. Rams WRs Robert Woods, Brandin Cooks and Cooper Kupp played seven full games together last season. The trio combined to handle a hefty 69 percent of the targets, with Woods (24 percent) just ahead of Cooks (23 percent) and Kupp (22 percent). Gurley still got his (15 percent), though Josh Reynolds (2 percent) and the tight ends (combined 12 percent) played lesser roles. Expect McVay's "11"-heavy offense to run through the wide receiver position again this season, especially Woods, who quietly has averaged a 24 percent target share in back-to-back seasons. All three receivers were in the top 11 in fantasy points during the seven games together.

24. Vikings RB Dalvin Cook has appeared in just 15 games through two seasons, due to a variety of injuries. During the 13 games in which Cook didn't miss a substantial chunk of time with an injury, he handled 61 percent of the team's designed runs (14.2 per game) and 14 percent of the targets (4.7 per game). That's RB1-level usage. Minnesota is committed to the run, as shown by its offensive playcalling following the promotion of Kevin Stefanski to offensive coordinator. Cook is positioned for a third-year breakout.

25. Vikings WR Adam Thielen enjoyed a 26 percent target share (8.7 per game) in 2017. That jumped to 28 percent (11.0 per game) through Week 14 last season. Once the team made the change to Stefanski, however, Thielen's playing time remained the same, but his usage took a major dip. During Weeks 15-17, Thielen handled a 15 percent share, totaling 12 targets in three games. Thielen is too good to become an afterthought, but with the Stefanski offense leaning more on Stefon Diggs, Kyle Rudolph and the run game, it seems likely that Thielen's targets will dip at least slightly in 2019.

26. The Patriots' offense is averaging 3.19 offensive touchdowns per regular-season game since Rob Gronkowski entered the league in 2010, including no fewer than 2.75 per game in a single season. Of course, Gronkowski missed 29 of 144 games in the span, and the splits are telling. With Gronkowski, the New England offense averaged 3.31 touchdowns per game (2.16 passing) and 288.6 passing yards per game while completing 66 percent of passes and posting a 7.9 YPA. Without Gronkowski, the numbers are 2.69 (1.72), 271.1, 61 percent and 7.0. For perspective, a 7.0 YPA would rank 22nd and a 7.9 YPA first in the league during the span. Tom Brady was held to five top-10 fantasy weeks last season (his fewest the past decade), and the New England offense is unlikely to pass much more often and figures to score less in 2019.

27. Here are Patriots WR Julian Edelman's target shares during weeks he has been active his past five seasons: 26, 26, 26, 29 and 26 percent. Any predictions for his 2019 rate? Edelman turns 33 this year, but especially following Rob Gronkowski's retirement, he's ticketed for another massive role. Durability is the drawback -- he has missed 45 of a possible 160 regular-season games (28 percent) since joining the team in 2009 -- but Edelman should flirt with WR1 numbers when active.

28. Patriots RB Rex Burkhead has appeared in 23 games in two seasons with the team. When active for 12 games in 2017, he played 25 percent of the snaps, handled 22 percent of the designed runs, ran a route on 24 percent of the team's pass plays and racked up a 9 percent target share. In 11 games last season, his marks were 27, 22, 23 and 8 percent, respectively. Burkhead has a defined role in this offense and is unlikely to disappear in 2019.

29. Speaking of the Patriots' backs, the trio of Burkhead, James White and Sony Michel played eight games together from Week 13 on last season (including playoffs). During those outings, Michel carried the ball 153 times, compared to 56 for Burkhead and 32 for White. White paced the trio with 55 targets, compared to 21 for Burkhead and four for Michel. On the season, New England called a pass on 79 percent of White's snaps, 55 percent of Burkhead's snaps and an absurdly low 24 percent of Michel's snaps. Although it's reasonable to expect second-year Michel's targets to rise a bit, it probably won't be by much with a similar dynamic expected.

30. Counting only weeks he was active, Saints TE Jared Cook has averaged target shares between 16 and 18 percent each of the past seven seasons. That's as consistent as you'll find in the NFL. Cook, who posted his first career top-12 fantasy campaign in Oakland last season, will be a no-doubter TE1 if he keeps his target share streak alive in 2019.

31. The 2018 season was two different stories when it came to the Sam Darnold/Robby Anderson battery. During Weeks 1-9, Anderson handled a 16 percent target share (5.25 per game) and was 59th at wide receiver in fantasy points. Darnold then missed four games and returned for the team's Week 14-17 games. During those four outings, he posted his four highest target shares of the season. One-third (or 10.0 per game) of Darnold's targets went his way, which helped Anderson to the seventh-most fantasy points during the span. Contract-year Anderson has massive upside if Darnold takes another step forward in 2019.

32. Eagles WR DeSean Jackson has appeared in 16 regular-season games twice in 11 NFL seasons. That's not good, but the silver lining is that he has appeared in at least 14 games in eight of those seasons. Jackson is a difference-maker when on the field, especially in non-PPR, but the 32-year-old will likely miss a game or two, which needs to be factored into his ranking.

33. Seahawks WR Tyler Lockett broke math by posting a ridiculous 57-965-10 line on 68 targets last season. His 14.2 yards per target (YPT) is second highest among 1,572 players who have recorded 50-plus targets in a single regular season since 2007 (Devery Henderson in 2008 is first at 14.4). Is Lockett's production sustainable? History says no chance. I looked at the top 10 in the category among those who saw 50-plus targets again the next season. All 10 took a sharp dive (see chart), and the average next-season YPT was 9.1. Lockett also exceeded his OTD by a league-high mark of 6.8, which means his expected touchdown total was closer to 3.2. Same as the aforementioned Chiefs offense, it's not a matter of if Lockett will regress in 2018 -- it's by how much.

34. Seahawks WR Doug Baldwin has averaged a target share of at least 22 percent each of the past five seasons. That includes 23 percent in 2018, which is the second-highest mark of his career. Injuries led to a slow start and a pair of missed games last season, but Baldwin finished strong. From Week 9 on, his share grew to 26 percent, and he was 18th at wide receiver in fantasy points. Baldwin is now 30 years old, and Seattle is committed to a run-heavy offense, but assuming he's healthy and ready for Week 1, it shouldn't be a surprise if Russell Wilson's top target posts his fourth top-20 fantasy campaign in five seasons.

35. Buccaneers WR Mike Evans averaged a target share of 26 percent as a rookie before bumping to 30 percent in 2015 and peaking at 31 percent in 2016. His usage has declined in recent seasons, with his share falling to 24 percent in 2017 before hitting a career low at 23 percent last season. It's possible Evans' usage bounces back under new head coach Bruce Arians, but Arians already said he thinks Chris Godwin could push for 100 receptions, and the team will need to find ways to feature O.J. Howard. Evans is still one of the game's top "big" receivers and has finished top-five in end zone targets each of the past three seasons, so he remains a fringe WR1.

36. Speaking of finding ways to get Howard the ball, that was a problem under the previous regime. Howard bailed out trusting fantasy owners by scoring touchdowns at an unsustainable rate (11 career TDs, 5.9 career OTD) but was limited to a weak 8 percent target share as a rookie and 11 percent share in 2018 when active. That's not going to cut it, and it's fair to be concerned here, as no Arians offense has seen a tight end unit eclipse a 20 percent target share the past 12 seasons. Howard has elite upside, but this offense could restrict him a bit, especially if Cameron Brate isn't traded (you up, New England?).

37. Titans TE Delanie Walker has registered a target share of at least 22 percent his past four seasons (excluding a 2018 season lost to injury after 38 snaps). The 34-year-old entered 2018 having posted three consecutive top-five fantasy seasons. Walker can actually afford a loss in targets and still see enough volume to allow a solid TE1 season. He has the look of a 2019 sleeper.

38. Redskins TE Jordan Reed handled a massive 23 percent target share (6.9 per game) during the 12 full games he played last season. That aligns exactly with his usage in the 2015 and 2016 seasons and is just above where he was in five "full" games in 2017 (21 percent). Despite scoring only two touchdowns in Washington's mess of an offense, Reed was fantasy's No. 8 tight end before he went down with an injury in Week 14 last season. He'll again be ticketed for a huge role in 2019, though his durability remains a massive concern. Reed has never played more than 14 games in a regular season and has missed 31 games in six seasons.

39. To wrap up, here is some regression analysis on leaguewide data from 2009 to 2019:

NOTE: R-squared is used to determine the accuracy of the statistical model used in the analysis. It is always between 0 and 1. The closer to 1 (the higher the number), the more accurate the model is, or to put it another way, the greater the R-squared, the better the correlation.

• What better predicts a team's playcalling: scheme/game plan or game script? Though both are relevant, the answer is the former. When comparing "preferred" pass rate with actual pass rate, we get an R-squared of .592. When comparing "expected" pass rate (which is based solely on score and time remaining in the game), the R squared is .1575.

• NFL teams have called pass 60.2 percent of the time, but 65.5 percent of touchdowns have been passes during the span. Is there correlation between the two? Not much. The R-squared is 0.26. Monitoring OTD will give you the best sense of a player's scoring opportunity.

• Comparing offensive snaps per game to offensive touchdowns plus field goal attempts (or "scoring opportunities") shows an R-squared of 0.216, and there's some slight correlation between plays and "expected" pass rate, which suggests that teams that aren't scoring points are running fewer plays. That makes sense, as poor offenses that can't score likely can't sustain drives.

• Establishing the run is how you win in the NFL, right? Nah. More in-depth studies have put that idea to bed, but even a simple experiment comparing game-flow-adjusted playcalling to scoring shows a positive correlation. The R-squared is 0.132 for touchdowns per game and 0.098 for scoring opportunities.

• This one is obvious, but let's quantify it: Teams that are trailing tend to score less. Comparing expected pass rate to scoring opportunities shows an R-squared of 0.539, and comparing it to touchdowns shows an R-squared of 0.450.