Statistically Significant: Average Separation Score

When we introduced the world to ASS (Average Separation Score) last year, we knew we had a killer stat that would revolutionize the game. But we never expected that one simple tweak could make it as or more predictive than virtually any other WR efficiency stat (as I’ll explain below). And even in its raw original form, ASS has proved even more useful than expected.

For example, the metric correctly indicated that Mike Evans (5th) wasn’t slowing down with age as many feared. He ultimately finished 9th-best at the position by FPG and scored the 3rd-most fantasy points (behind only Puka Nacua and Justin Jefferson) over the final six weeks of the fantasy season.

Tee Higgins and Jerry Jeudy stood out as players who’d strangely earned fewer targets per route run (TPRR) than their ASS would have suggested. Higgins went on to see roughly a ~13% bump in his TPRR and finished as the WR6 by FPG, while Jeudy had the best statistical season of his career and finished inside the top-24.

But admittedly, none of these three players accomplished their feats through exemplary separation alone; ASS merely suggested they’d take advantage of an opportunity if presented with one. Evans was a benefactor of Chris Godwin’s injury, Jeudy needed some Jameis Winston magic, and Higgins (along with the rest of the Bengals’ offense) benefited from the extra pass volume necessitated by an unexpectedly horrific defense. Fantasy football is often just as much about these external factors as a player’s inherent skill. And even then, separation is only one element of a WR’s skill — catch radius, body control, ball skills, contested catch ability, and after-catch explosiveness can “earn” a player targets even without great separation.

So, how predictive is ASS of fantasy scoring by itself? What limitations does it have, and how can we work around them to more intelligently get an edge in fantasy football? By the end of this article, I’ll have answered these questions and uncovered several underrated players to target in your 2025 fantasy football drafts.

As always, feel free to scroll down to the player blurbs if that’s all you’re here for.

How Predictive Is ASS, Actually?

Starting on a per-route basis — where the external factors we listed are largely controlled away — ASS is incredibly predictive of fantasy points.

Separation Score	FP/RR	TPRR
-2	0.02	14.0%
-1	0.09	19.8%
0	0.23	15.0%
1	0.82	40.9%
2	1.71	55.3%
3	2.53	59.8%

Over the past three seasons, a positively graded route (1, 2, or 3 separation score) has been worth 4.39x more fantasy points than a negative or neutrally graded route. They were also 2.78x more likely to result in a target. Furthermore, the degree of separation matters, speaking to ASS’s relationship to fantasy scoring across the entire scale. “Perfect” separation (3) is worth nearly 11.0x more fantasy points than a neutral (0) route (although players were targeted on only ~60% of their perfect routes).

But while the above demonstrates that ASS is great at telling you how many fantasy points we can expect a WR to score on each route, it doesn’t tell you how many routes a WR will run. Referring to our examples above, Tee Higgins ran 21% more routes per game in 2024 than the prior year, going from the 48th-most to the 6th-most. Jerry Jeudy saw an even larger 46% increase in route volume, going from 61st to 2nd. These players’ ASS suggested they’d take advantage of any positive situational changes, such as Higgins’ better health and horrible defense, or Jeudy’s higher-volume offense leading to more routes.

Other examples of situational factors that ASS deliberately looks past are designed targets (covered here), QB quality (covered extensively in the player blurbs at the bottom of this article), and target competition. We can see the latter in action using another of our charted stats, first-read targets.^[1]

Separation is further rewarded when the QB throws to his first read. A positive separation score is worth 6.3x the fantasy points of a neutral or negative score on first-read plays, but that drops to just 2.0x on all other plays.

Separation Score	FP/RR when QB throws to the first read	FP/RR on all other plays
Neutral or Negative (-2, -1, 0)	0.21	0.36
Positive (1, 2, 3)	1.32	0.73

Although the way we chart first reads makes this a bit more complicated than I’m stating,^[2] in short, the more often a WR is on his QB’s first read, the more reliably his separation abilities will translate to targets and fantasy points.

This is why ASS is agnostic of target competition. Mike Evans can separate all he wants, but he won’t score any fantasy points on that play if Chris Godwin also separated as the first read on the other side of the field. But Evans’ excellent ASS provided us a preview of exactly what would happen once that target competition went away: in six games with Godwin, he averaged 14.3 FPG (~WR22) on a 25.2% first-read target share. In seven games without him, that jumped to 20.7 FPG (~WR2) on a 32.5% first-read target share. A lesser WR with a worse ASS likely wouldn’t have taken as big an advantage of this situation.^[3]

Last offseason, we’d already found ASS to be more stable and predictive than competitors like ESPN’s Open Score (despite that metric — at least as of late 2022 — incorporating a bonus for being targeted rather than purely charted observations). Unlike Next Gen Stats, we chart separation on every route (not just ones where the player is targeted), and don’t give credit to players for being “open” on screens. Still, the confounding external factors we’ve discussed mean that despite those promising initial results, ASS always had an uphill battle to be directly useful for fantasy from season to season.

Counting stats and production-based metrics (receiving yards, target share, yards per route run, etc.) have inherent advantages in that they automatically bake in whatever contextual factors (route volume, target competition, QB quality, offensive competency, etc.) led to that production. Since most of those factors will remain roughly the same for most players from season to season, those counting stats are broadly more predictive.

On its own, ASS doesn't come anywhere close to these other metrics (see below). However, once we introduce a single key variable, it becomes highly predictive. (I'm getting there; just give me a few more paragraphs.)

Can We Improve ASS For Season-Long Fantasy Football Formats?

As last year’s examples of Evans, Higgins, and Jeudy demonstrate, ASS is best used to provide a rough “skill check” for a WR. We often can’t predict the chaos of injuries, game script, and league transactions before an NFL season, but ASS provides a baseline for how a player might interact with that randomness.

With that in mind, the way to improve ASS and make it more useful for fantasy football becomes clear. Rather than combining it with counting stats to shove back in a few of the key external factors our film charters have worked so hard to strip out (i.e., without “cheating” like ESPN does, no offense), I wanted to augment its ability to identify players best-positioned to capitalize on positive opportunities — whether foreseeable (like getting an offseason QB upgrade) or unforeseeable (like a mid-season injury to a teammate).

After a lot of testing and tinkering, I realized the best way to accomplish this was by smashing ASS together with Route Share. After all, the more often a WR is on the field and running a route, the more opportunities they’ll have to take advantage of the separation they create, intensifying the fantasy football impact of any lucky breaks they catch. It also further rewards WRs who can maintain their separation results across a wider range of routes, game situations, and defensive matchups, rather than only a handful of spots where a coach believes a part-time player is best deployed.^[4]

To avoid overfitting while working with a limited 3-year sample of data, I created a model based on each WR’s 2022 ASS and Route Share in an attempt to predict their FPG in 2023. Then, I tested the model by evaluating its ability to utilize 2023’s ASS and Route Shares to predict each player’s FPG in 2024.

I call this metric “Playing-time Adjusted Separation Score” (PASS). PASS was more predictive than industry-favorite efficiency metrics, such as YPRR and 1D/RR, without relying on any counting statistics, raw volume, or actual production. (I qualify this statement and provide some additional context within the footnote at the end of this section.)

This makes PASS perfect for evaluating the upside of WRs switching teams, getting new QBs, or otherwise experiencing a new situation. Overall offensive volume, QB accuracy, previous target competition, and designed targets should all have no bearing on PASS, unlike any other available metric. All that matters is a player’s inherent ability to get on the field and separate.^[5]

To help readers gain as big an edge as possible, I fed all our available years of ASS and fantasy results (2022-2024) into the PASS model and ran the results for the 2024 season.^[6] Let’s see what we can glean for your 2025 fantasy drafts.

2024 PASS Leaders and 2025 Breakouts

15 of last year’s top 24 WRs by FPG also ranked top 24 in PASS. It’s no surprise to see superstars like Ja’Marr Chase, A.J. Brown, Malik Nabers, Justin Jefferson, Amon-Ra St. Brown, Ladd McConkey, Mike Evans, Tee Higgins, Brian Thomas, Davante Adams, and Nico Collins here; this is a welcome confirmation of their outstanding skills, and again, this metric identified them without including any measures of on-field production.

Oh, and if I end up being wrong about my Jameson Williams fade in 2025, this is where I’ll point myself to pay even greater attention to in future seasons.

But with those players acknowledged, I want to focus primarily on the more surprising names.

• As his top-5 PASS ranking implies, might we be talking about DeVonta Smith as one of the NFL’s best receivers if not for the Eagles’ 31st-ranked dropback volume (535) and the imposing presence of A.J. Brown, who last year commanded the league’s 6th-highest TPRR (29%)? I believe so. In the only two games he’s played without Brown since 2022, Smith averages 17.8 FPG (~WR7) on a 29.4% target share (~WR3). And we may even get a similarly powered-up Smith in 2025 if Dallas Goedert is indeed traded: he averaged 17.5 FPG (albeit on just 6.1 targets per game) across seven contests with Brown but without Goedert last season. Smith goes as just the ~WR28 across most platforms, but he has outperformed that ranking in every season since his rookie year and has among the best contingent upside of any receiver in the league.

• Perhaps no receiver has been more hampered by external factors than Garrett Wilson over his career. He was largely forced to play with Zach Wilson for his first two seasons, the NFL’s least accurate passer of the current decade. Then, over the first six weeks of 2024, he led the NFL in targets (66) while averaging 21.1 XFP/G, commanding overall WR1-levels of volume with Aaron Rodgers. But from that point on, he was forced to compete with fellow top-24 PASS receiver and Rodgers’ best friend, Davante Adams. Wilson’s first-read target share fell to just 29.3% (22nd) — a metric by which he’d ranked top-10 in all of his previous seasons — ultimately culminating in heated late-season exchanges on the sideline. Wilson is the exact sort of player you’d expect to pop in a “skill check” metric compared to his counting stats, but even with Adams gone, I’m not sure he’ll catch a break in 2025. New Jets QB Justin Fields is Zach Wilson-esque in his ability to waste dropbacks on sacks, throwaways, and inaccurate throws.

• Sam Darnold’s less-than-stellar showings in the two fantasy-friendliness metrics above highlight the chance J.J. McCarthy (Brett Whitefield’s QB1 in last year’s draft class) represents an upgrade for Jordan Addison. Excluding Darnold’s Week 18 primetime meltdown that cost the Vikings the #1 seed and likely ensured the team would move on from him, Addison averaged 15.8 FPG (~WR14) in his healthy games, or 16.3 FPG in games where all of Addison, Justin Jefferson, and T.J. Hockenson were available. Even after factoring in his 3-game suspension, Addison’s ~WR37 ADP doesn’t make much sense. Though his fantasy success so far has been partially attributable to elevated TD production, that could be here to stay in an offense that’s ranked top-5 in red zone pass rate during every year of Kevin O’Connell’s tenure and that fittingly fed Addison 0.8 end zone targets per game in 2024 (13th-most). Addison produced this way despite the Vikings ranking only league-average by total dropbacks (626, by far the fewest of the O’Connell era) due to passing only 56.1% of the time in the 2nd half (8th-least) — a trend that would likely reverse if the Vikings have any less than a top-3 defense in 2025. All told, there are plenty of outs for Addison’s situation to get better this year, further accentuating the inherent separation skill his top-24 PASS rating reflects. If the already-banged-up Justin Jefferson misses time at any point this year, Addison has top-12 upside in those games.

• Zay Flowers has ranked top-6 in PASS during both his rookie and sophomore seasons, but failed to crack the top-30 by FPG in either. Separation-related Lamar Jackson MVP debates aside, the Ravens have attempted the fewest passes in the NFL over the past two seasons (971) and threw the ball just 42.2% of the time while leading in 2024 (2nd-least). Through the first 10 weeks of the season (before the Ravens’ defense began leading the NFL in every category), Flowers averaged 15.3 FPG (~WR15) in his healthy games and exceeded 18.0 PPR points a whopping five times, fewer than only Justin Jefferson and Amon-Ra St. Brown. There’s also an argument that a smaller WR like Flowers will naturally have trouble performing up to his separation numbers and holding up physically through an NFL season^[7], but I mostly fall on the side of him being a victim of positive game script and his team’s success at running the ball. That won’t necessarily be different in 2025, but Flowers’ ~WR31 ADP isn’t a horrible price to get access to any shootouts the Ravens might find themselves in (at least on Underdog — he goes much higher than I’m comfortable with on ESPN and Yahoo!).

• Rashod Bateman was similarly disadvantaged by gamescript, but his separation numbers don’t indicate as much fantasy football upside to me as Flowers’. Bateman’s 15.0 aDOT ranked 5th-highest among WRs with 60+ targets, indicating frequent downfield usage and very few fantasy-friendly screen targets (Flowers has 50 designed targets over the past two seasons to Bateman’s zero). Downfield routes generally mean fewer catchable targets, worse per-route efficiency, and less weekly consistency in fantasy football.^[8] That’s a huge problem in managed formats, but Bateman is one of the final useful WRs before the position’s cliff at his ~WR57 Underdog ADP.

• This is also a concern for Marvin Harrison, whose route usage and 2025 outlook I broke down in more detail here and here. However, unlike with Bateman, we can hope for a Year 2 breakout by a generational WR prospect. And it’s an encouraging sign that Harrison ranked top-24 in PASS during his disappointing rookie season, compared to just 40th in a more situation-affected stat like YPRR (1.74). But even if Harrison’s usage and route tree changes, his situation won’t; Trey McBride just became the NFL’s highest-paid TE, and Kyler Murray ranked just 21st in accuracy rate on throws traveling 10+ yards downfield (36.8%). There’s obvious upside here, but I’m still not particularly excited to pay Harrison’s ~WR18 price tag in redraft. But if Harrison underperforms for fantasy again in 2025 while maintaining a strong PASS rating, I suspect I’ll be interested at a reduced cost in dynasty formats next offseason; Murray has no guaranteed money on his contract after 2026.

• Brandon Aiyuk may not be ready to play until midseason and is far from a lock to make a major fantasy impact in 2025, but up until he tore his ACL and MCL, he was separating better than any WR in the league. However, his chemistry with Brock Purdy likely suffered from missing the entire preseason due to his contract holdout, resulting in just 8.9 FPG on a 19.9% target share (~WR40). This wasn’t unprecedented nor unique: fellow 2024 preseason holdout CeeDee Lamb averaged just a 21.8% target share (~WR34) over the first six weeks, a 25% reduction from his 2023 number. In fact, all three WRs who held out in 2024 — Aiyuk, Lamb, and Ja’Marr Chase — were targeted much less often on a per-route basis over the first six weeks compared to the rest of the season, regardless of their separation numbers (below). Again, it’s likely a moot point for 2025 due to Aiyuk’s injury, but I found this explanation much more convincing than the idea that separation isn’t as heavily rewarded in Kyle Shanahan’s scheme.^[9]

• Much like many of the other “surprising” names at the top of the PASS list, Romeo Doubs was stuck on a low-volume pass offense in 2024, the Packers having dropped back the fewest times of any team (521). He’s also dealt with stronger target competition for much of his career than he’s been given credit for; while Dontayvion Wicks, Christian Watson, and Jayden Reed don’t score well in PASS nor possess great counting stats due to their lower route shares, the trio ranked 3rd-best, 6th-best, and 50th-best in unadjusted ASS last season. Doubs must be doing something right (mostly In and Out routes) to be the only WR of this group entrusted with a true full-time role. I’d be pretty intrigued if Doubs ever ended up on a different team that would ask more of him, but it doesn’t seem like that will happen in 2025. While Watson will likely miss much of the season, the combination of Wicks and Round 1 rookie Matthew Golden taking over Watson’s routes on the other side likely won’t be a big enough boost to Doubs. (Doubs averages just 11.3 FPG on 6.4 targets per game across nine contests without Watson over the past two seasons.) This offense is built to maximize separation, but to minimize the fantasy production of its individual WRs.