We are all searching for an edge in our fantasy leagues, and we all go about it differently. Camp reports, coach talk, projections, ADP, etc., each carry varying weights of importance. Odds are, this isn’t the first projection model you’ve seen, nor will it be the last. So what makes this different?
My projection models were at the root of my FantasyPros WR draft rankings that took 1st overall last year. Projection models form a base for understanding the range of “likely” outcomes for a player, and my models are what I put the most weight in when determining my final ranking for the year.
Using two independent model projections with varying – yet still correlative – inputs have shown to improve accuracy by combining them together. This levels the peaks of valleys of each modeling technique or amplifies when a player really stands out above the crowd.
Let’s start with wide receivers first and go through the two models I am using. They blend both individual projections with team passing projections.
Join our Discord to Chat with Experts and FantasyPros Subscribers
Multiple Linear Regression (MLR) Model
A staple among most projection models, MLR identifies one-to-one trends among each stat category to fantasy PPG. I tried to select only the variables that were stable year over year and carried p-value significance for the MLR study. For wide receivers, they were the following categories:
- Targets per Game
- Receiving Yards Per Game
- Touchdowns Per Game
- Completions Over Expected (COE)
- Derived from Completion %, aDOT exp completion %, and QB on-target %
This is rather basic apart from COE as it only uses four variables, but that’s okay to start. These stats carry some of the highest correlations to next year’s fantasy points per game, however unexciting they may be. We are building a stable base, and this will help balance out the second model.
Signature Efficiency Dampening and Team Passing Model
This one gets a little complex. I’ll start with the categories used and then explain how they are used to output a projection:
- Team Passing Volume
- Derived from Vegas Win Totals, Coaching Tendencies, and League Regression
- WR Target Share
- Player specific targets, target share, and team volume
- Efficiency Dampening
- Combines individual efficiency rates with league average regression
- Yards per receptions
- Yards per touchdown
- Also factors in Vegas team odds and Team Yardage Scoring Efficiency
- Completion %
- Combines individual efficiency rates with league average regression
Team passing volume is at the root of these projections. Perfect total passing projections are impossible, but I have liked the range these three factors give for pass attempts expected for each team over the years. Below is an example of the league regression variable I use, which is one of the three factors:
Next up is target share. This one is a little trickier when personnel changes over, but again I am balancing an individual’s target share with regression on their new expected team totals. This variable has the most variance out of any factor in the model, but the balance between the first model’s targets per game and the new team projected target share gives another balancing act. Having two models source two different sources for targets helps create a fair balance.
Finally is my favorite part, efficiency dampening. This uses an average of player-specific efficiencies, league-average rates for their similar ranged players, and team trends.
Let’s use an example. Julio Jones had a yards per reception of 15.1 last year. This model gives his individual rate a majority favor but then averages it against the typical WR rate of 13.5 yards per reception. So while Julio is routinely a deeper threat in the league, this helps flatten out variance in things like yards after catch and average depth of target, or if the coaching scheme or route tree changes up a bit.
Again, I am aiming for balance and regression while still crediting the WR himself for past efficiency marks. Another example is a guy like DJ Moore, who routinely has low touchdown rates compared to yards and receptions, so while I expect a slight bump this year, it still projects him being below league average for his usage based on his proven history and the team’s projected scoring rate this year.
Now that we have an understanding of the process let’s get to the projections.
2021 WR Model Projections
For our 2021 projections, I’ll use both the model projections and expert consensus rankings (ECR) from FantasyPros to break down the top values and highest risk players among the top-50 ECR WRs. Model projected PPG, expert rankings (ECR), and Final Rank (approximate average between the model and ECR) are listed below:
Take some time to compare the projections versus their consensus ECR. I won’t be able to cover them all, but here are some of the best and worst values from the model.
Top Values
Tyler Lockett (WR – SEA) – WR14 vs. WR21 ECR
Known for his boom or bust reputation, Lockett was a blessing and a curse to fantasy owners last season. In fact, he had only four games with over 70 yards; however, the model loves his target volume and expects his volatile per game variance to level out more this year at a steadier efficiency clip.
Brandin Cooks (WR – HOU) – WR23 vs. WR37 ECR
I seem to write about Cooks every single year, and he beats out ADP like clockwork nearly every single year. Yes, it is devastating to no longer have Watson throwing him the ball – it is all but assumed he will be elsewhere or suspended – but he might be the best and only top talent on this entire Texans offense. What’s stopping him from a 25% target share? Anthony Miller? Nico Collins? Feels doubtful. Whether it’s Tyrod Taylor or rookie Davis Mills, he will be their most reliable and trustworthy receiver.
Corey Davis (WR – NYJ) – WR28 vs. WR43 ECR
Davis nearly matched fellow receiver and fantasy darling AJ Brown in targets and yards per game last year, but you wouldn’t know that looking at their ADP differential. Davis goes to a completely rebuilt Jets team with new coaches, a new quarterback, and a team projected to increase its passing volume by nearly four attempts per game as per my team study above. They paid Davis to be the alpha there for now, and for a guy that went shoulder to shoulder with one of the top receivers in the league last year, I’m betting on him being able to deliver.
Other notables: Will Fuller, Antonio Brown, and Emmanuel Sanders
Highest Risks
Mike Evans (WR – TB) – WR21 vs. WR17 ECR
Evans and Adam Thielen were the two most touchdown-dependent receivers in the league last year. He scored on an absurd 18.6% of his receptions. He has always been a tremendous red zone threat, but that is way too high of a rate for the model to expect a repeat of. A full season of Antonio Brown, OJ Howard, and Giovanni Bernard should cut ever so slightly into the target share as well, but that’s not nearly as big of a concern as needing him to score nearly every game to pay off at cost. His 4.4 receptions per game was the second-lowest mark among top-24 WRs last year.
CeeDee Lamb (WR – DAL) – WR24 vs. WR14 ECR
Lamb remains a tremendous prospect, but expectations are spiraling upwards for the second-year wideout. The Cowboys ran the second-highest number of plays in the league last season, along with the second-highest pass attempts per game. If Lamb is going to hit a WR14 finish, they will both need to keep up that pace, and he will need to steal a healthy amount of targets from Amari Cooper and Michael Gallup. The return of Dak Prescott should increase the quality of their targets, but the model believes another top-24 finish is a much more likely outcome than top-12.
Kenny Golladay (WR – NYG) – WR33 vs. WR23 ECR
Golladay goes to a team with not only much more competition for targets than the Lions had – Sterling Shepard, Darius Slayton, and Saquon Barkley – but a significant downgrade in quarterback from Matthew Stafford to Daniel Jones. Giants were 16th in the league in passing rate last year, and that was without Barkley to hand the ball to. On the bright side, I don’t think Golladay will ever leave the field in their offense. My team model has them as the second-lowest scoring team next to the Jets. One could say he is in a similar situation to Corey Davis, but a big difference being the projected pass attempts per game for each team in 35.2 for the Jets and 32.6 for the Giants.
Other notables: Laviska Shenault (WR – JAC), Tee Higgins (WR – CIN)
Thanks for reading, and stay golden! If you like what you learned, follow me @DavidZach16 for more interesting stats and analysis throughout the year.
Can you draft the perfect 2020 team? Try our Perfect Draft Game
Subscribe: Apple Podcasts | Spotify | Google Podcasts | Stitcher | SoundCloud | iHeartRadio
If you want to dive deeper into fantasy football, be sure to check out our award-winning slate of Fantasy Football Tools as you navigate your season. From our Start/Sit Assistant – which provides your optimal lineup, based on accurate consensus projections – to our Waiver Wire Assistant – that allows you to quickly see which available players will improve your team, and by how much – we’ve got you covered this fantasy football season.
David Zach is a featured writer at FantasyPros. For more from David, check out his archive and follow him @DavidZach16.