Projecting the Pros on DraftKings: Week 10

A guide to playing roster percentages like the pros.

NOTE: The pro projections below are first posted on Thursday, then updated on Saturday afternoon. On Thursday, commentary will be limited to early impressions. A more-detailed discussion of specific players will be added on Saturday afternoon.

Welcome back for Week 10 of Footballguys' new feature that uses a statistical model to project the roster percentages of "professional" tournament players on Draftkings. If you're curious about the full methodological details of the model, click here. The most important detail to keep in mind, however, is that the "professionals" in question were identified using a clustering algorithm and can be characterized by the following statement: Compared to the rest of the population, these 205 pros play more, they win more, and they do both consistently.

Quarterbacks

Below is the table for quarterbacks. In this and all subsequent tables, "Overall" is Steve Buzzard's projection of roster percentage. "Pros" is the model's projection of roster percentage for Draftkings pros only.

Early Impressions

If Jared Goff seems like he's at or near the top of the table every week, that's because he is. And once again, it's due to the gaudy roster percentage projection for his wide receiver corps (36.1% combined). One surprise is Alex Smith being projected to appear in fewer pro lineups than overall despite having one of the best quarterback matchups of the week. The reason is because Smith's the anti-Goff right now, with only an 8.1% aggregate roster percentage projection for his wideouts.

SATURDAY UPDATE: In a stunning turn of events, Goff isn't even in the Top 3 anymore. Based on related changes at other positions, it appears that the public has moved in the direction of Rams-Seahawks being lower-scoring and more lopsided than initially thought. Meanwhile, Smith's standing has hardly changed since Thursday.

Aaron Rodgers ($6,400, 8.6% Overall Roster Percentage, 9.4% Pro Roster Percentage)

Rodgers has the same salary for a home game against the Dolphins as he did for last week's road game against the Patriots (::thinking emoji::). Miami ranks eighth in Draftkings points allowed to opposing quarterbacks, but that's a tad misleading. They're getting a big boost from having the second-most interceptions in the league. Picking on the likes of Sam Darnold, Blaine Gabbert, and Derek Carr will do that for you. At last check, Rodgers is better than those guys and has thrown only 1 interception all season.

With respect to the pro projection model, Rodgers benefits from having what figures to be the highest-rostered wide receiver corps on Sunday's Main Slate (30.0% combined). In addition, the aforementioned Dolphins defense is projected to be one of the lowest-rostered (1.1%).

Ryan Fitzpatrick ($5,900, 9.4%, 10.2%)

After coming out of the gate hot this season, Washington has allowed five consecutive quarterbacks to achieve at least 3.6x value, including last week's 5.6x performance by Matt Ryan. The softness of this matchup is one reason why Fitzpatrick's point projection is so high (21.7), and said point projection is why he Ranks No. 1 among Main Slate quarterbacks with a 41.1% probability of achieving 4x value. The model also gives Fitzpatrick a boost because Tampa Bay's wide receiver corps combines for 24.8% projected rostership, which ranks fourth.

Running Backs

Below is the table for running backs:

Early Impressions

Todd Gurley (for the umpteenth time) is an intriguing high-salary option, while Isaiah Crowell is an intriguing low-salary option. Of course, "fool me thrice..." If that sentence gave you déjà vu, it's because it's the thing you read in this space last week. And with that bit of jinxing done, this is no doubt the week that Crowell goes off for 35 points.

SATURDAY UPDATE: For the second straight week, the gap between Gurley's overall projection and pro projection closed significantly between Thursday and Saturday. Unlike last week, this time it's almost completely disappeared. Meanwhile, Crowell's projection has decreased, but his overall/pro gap remains the same.

Dion Lewis ($4,800, 28.6%, 37.3%)

Lewis was a favorite of the model on Thursday, but the situation has gotten even more favorable since then. At 31.4%, Lewis now has the second-highest 4x value probability among running backs. Typically, anything over 30% puts a running back into must-roster status. What's somewhat surprising about the model's projection is that Lewis is merely middling per the other two running back factors. New England's defense only has the seventh-lowest overall projection on the slate (6.4%), and Tennessee's wide receiver corps actually combines for the eighth-highest projection (16.5%), the latter of which is a particularly negative indicator.

Elijah McGuire ($3,400, 7.1%, 8.3%)

McGuire is the only running back ahead of Lewis in the 4x value probability rankings, at 34.7%. Unlike Lewis, other model factors are also favorable. First, Jets wide receivers combine for the seventh-lowest projected rostership (7.1%). Second, Buffalo has the seventh-lowest defensive roster projection. OK, so perhaps those rankings are more "middling" than "favorable," what with 22 teams on the Sunday Main slate, but at least they're both pointed in the correct direction. In other words, there's a coherent story being told here. But of course, if you want to pivot away from McGuire, there's always the aforementioned Isaiah Crowell -- if your stomach can handle it.

Wide Receivers

Below is the table for wide receivers:

Early Impressions

Michael Thomas is this week's "chalk that pros are likely to roster more often than the overall public." Last week it was Courtland Sutton, which didn't work out well as he only achieved 2.2x value on his salary. One phenomenon to dig deeper into during the offseason is the seeming trend of underperformance by low-salary chalk. As for Thomas, who is not in that category, the pro projection model favors him due to Cincinnati's defense being projected to appear in less than one percent of GPP lineups. If they're not worth rostering, then Thomas should be rostered more.

SATURDAY UPDATE: Almost nothing has changed for Thomas since Thursday. His projection is slightly higher, but the 5.8% overall/pro gap is exactly the same. Giddyup!

Corey Davis ($4,500, 16.5%, 18.6%)

Gonna be honest; not completely sold on this one because Davis has proven to be more of a steady-as-she-goes, cash game asset than a high-variance GPP asset. That said, the model spits out what the model spits out. And what the model sees is a wide receiver whose 28.1% probability of achieving 4x value ranks second on the slate.

Maurice Harris ($3,900, 7.4%, 7.7%)

Now we're talkin'! Harris is the high-variance GPP play we're looking for, as evidenced by four games of nothing followed by a 25.4-point eruption last week against Atlanta. And now, with Jamison Crowder out once again, Harris gets to enjoy a full game's worth of targets against the Buccaneers, who still haven't figured out that you have to cover the slot. At 25.8%, Harris has the fourth-highest 4x value probability among wide receivers.

Marvin Jones Jr ($5,500, 6.3%, 6.7%)

With a bad matchup coming on the heels of a poor offensive performance by the Lions in their first game of the post-Tate era, recency bias may very well result in Jones flying under the radar this week. Indeed, his overall roster percentage projection seems a little bit high. One man's trash is another man's treasure, and Jones has booty of a third-ranked probability of achieving 4x value (26.3%). If it gets late early in the Chicago-Detroit game, Jones will likely see more than enough targets to pay off his salary. Or it'll end up being a complete wipe out. But hey, that kind of low-floor, high-ceiling uncertainty is what we want in GPPs. #EmbraceTheVariance

Tight Ends

Below is the table for tight ends:

Early Impressions

Despite going against Arizona, Travis Kelce has as narrow a margin between overall and pro rostership projection as he's had in weeks. This is mostly due to his salary now standing at $7,000, which is $400 more than last week and only $100 less than his season-high in Week 6. Of course, what's weird is that his Week 6 opponent was New England, whereas his Week 10 opponent is Arizona.

SATURDAY UPDATE: The gap between Kelce's overall and pro projections has widened significantly. The change is due to the projected roster percentage for Arizona's defense decreasing slightly, from 0.9% to 0.7%. That may not sound like much, and it isn't, but every little bit helps, especially at the extremes of the table.

Vernon Davis ($2,800, 1.6%, 3.1%)

Davis is the way to go if you want to completely punt at the tight end position this week. The main factor in Davis' pro roster projection is his No. 1 ranking among tight ends with respect to probability of achieving 4x value (24.7%). This is what makes him favored by the model, whereas Jordan Reed is not favored. That said, Tampa Bay allows the second-most Draftkings points to opposing tight ends and has surrendered a touchdown in five consecutive games when facing a player you've ever heard of. In other words, there's plenty to eat here. If you want massive salary savings, roster Davis. If you don't, using Reed isn't the worst option as the model is somewhat "splitting the baby" regarding Washington's tight ends.

Defenses

Below is the table for defenses:

Early Impressions

"Maximize exposure to defenses facing Buffalo" is now on par with "maximize exposure to Todd Gurley" and "maximize exposure to Adam Thielen." so it's no surprise that pros will be maximizing their exposure to the New York Jets.

SATURDAY UPDATE: The Jets remain atop the table, but now have company consuming that rarified air. None of the underlying model metrics moved much for the Rams. Their rise from Thursday is simply a byproduct of Buzzard significantly increasing their overall projection.

Indianapolis Colts ($2,800, 3.3%, 3.6%)

You could do worse than rostering a defense hosting a Blake Bortles-led offense that allows the third-most Draftkings points per game to opposing defenses. In fact, Bortles' low projected roster percentage (1.8%) is the main factor why the model projects pros to roster the Colts defense more frequently than the overall public.