We just finished the our veteran yearly model — new data pipeline, refreshed feature set (age curves, decline flags, 3-year trend, the works), fully retrained. First thing we do every time we ship a new version: run it, sort by projected PPR points, and see who the model is in love with. Here's who came out on top at each position, and the reasoning behind it as best we can explain it.
Quick disclaimer since it always comes up: this is a projection model, not gospel. It's trained on nflverse play-by-play going back to 2000, and it's scoring based on standard PPR (6pt passing TDs). Rookies with no veteran track record get less signal to work with, so take those with an extra grain of salt.
---
## Quarterbacks
**1. Josh Allen — 361.8 pts**
No surprise here. The model doesn't need fancy efficiency metrics to figure out Allen is special — he's the rare QB who racks up rushing value on top of an already-elite passing profile, and that dual-threat floor is exactly what separates the QB1 tier from everyone else in a PPR-style scoring system. Volume plus rushing equity is a cheat code the model rewards heavily.
**2. Patrick Mahomes — 352.6 pts**
Still Mahomes. Age is creeping into the picture as a feature now (he's on the wrong side of 30), but the model isn't seeing enough decline in the underlying passing efficiency yet to knock him down the board. Volume and TD rate carry him.
**3. Bo Nix — 346.3 pts**
This is the one that jumps out. Two years of NFL experience and already top-3. The model is picking up on a real developmental trajectory here — the offense is leaning into his mobility, and unlike a lot of young QBs he's not just a system game manager, he's producing rushing juice too. Book that as a "the model is buying the breakout" projection.
**4. Jalen Hurts — 340.8 pts**
Same story as always with Hurts — the passing numbers alone wouldn't put him here, but the rushing/goal-line role is worth a ton in a per-stat model like ours. That's a structural, scheme-driven floor, not noise.
**5. Jared Goff / Joe Burrow — 327.3 pts (tie)**
Two very different paths to the same number. Goff's model score is being propped up almost entirely by efficiency and volume in a clean pocket-passing offense — no rushing equity needed. Burrow's the classic "target volume + explosive-play upside" profile when he's on the field.
---
## Running Backs
**1. Jahmyr Gibbs — 265.6 pts**
The model loves what Gibbs represents structurally: real receiving work stacked on top of the rushing role. That dual usage is the single biggest lever in our RB feature set — a back who's purely a rusher caps out fast, but pass-game involvement changes the whole shape of the projection.
**2. Bijan Robinson — 256.1 pts**
Bijan's basically the same archetype as Gibbs — three-down back who's fully integrated into the passing game — and the model treats that role the same way regardless of name recognition. This isn't the model discovering something new, it's confirming what already became obvious on the field in 2025.
**3. Devon Achane — 242.3 pts**
Efficiency machine. Achane doesn't need bell-cow touches to profile well because his per-touch value (rushing and receiving both) is high enough that volume becomes a secondary concern. When a back is this efficient, the model doesn't punish him for a lighter workload.
**4. Ashton Jeanty — 233.4 pts**
**5. Christian McCaffrey — 226.6 pts**
CMC's age (30) and workload history are exactly the kind of thing our new decline-flag features are built to catch, and it's still not enough to push him out of the top five — that tells you how dominant the receiving role has to be to offset it. If he ever loses the passing-game role, this number falls off a cliff.
---
## Wide Receivers
**1. Amon-Ra St. Brown — 266.1 pts**
Target share is king for this position in our model, and St. Brown's is about as high and stable as it gets. Efficient, high-volume, low-variance — the exact profile the model is designed to reward at WR.
**2. Ja'Marr Chase — 262.6 pts**
Chase pairs elite target share with real explosive-play equity (yards after catch, big-play rate), which is the one-two punch that usually separates a true WR1 from a high-target possession guy.
**3. Puka Nacua — 246.3 pts**
Two years in and already a target-share monster in an offense that spreads the ball around by design — that's the model picking up on real, sustained role security, not a one-year fluke.
**4. Ladd McConkey — 240.8 pts**
Sneaky one for the model to like this much. It's rewarding the efficiency profile here — he's not necessarily leading his offense in raw volume, but the per-target production is strong enough to carry the projection.
**5. Jaxon Smith-Njigba — 235.4 pts**
Big third-year jump. The trend features are exactly what's driving this — the model is picking up on real year-over-year growth in role and production, not just projecting last year's raw stat line forward.
---
## Tight Ends
**1. Trey McBride — 198.3 pts**
McBride's basically playing the target-share game that separates WR1s from WR2s, except at a position where that level of receiving volume almost never happens. That gap between "TE" and "true focal point of the passing game" is exactly why he's this far ahead of the field.
**2. Travis Kelce — 187.7 pts**
Age is a real drag on this profile now (37, 13 years in), but the receiving role and offensive environment are still good enough to overcome it in the model's eyes. This is a name where you should watch the decline features more than the header number — it won't take much slippage to knock him down next cycle.
**3. Kyle Pitts — 172.1 pts**
Perpetual "the talent and target profile say this should be higher" case. The model's still buying the long-term equity even though the week-to-week role has been inconsistent.
**4. George Kittle — 172.0 pts**
Efficiency and offensive scheme carry Kittle more than raw target volume — he does a ton with what he gets, and the model picks up on that per-target value even when the target count itself isn't gaudy.
**5. Colston Loveland — 156.2 pts**
Rookie, so same caveat as Jeanty above — this is mostly a role/draft-capital-based estimate since there's no veteran trend data to lean on yet. Worth checking back on once we've got a real season of data behind him.
---
That's the full rundown. As always — this is one model's read on volume, efficiency, and role heading into next season, not a lock. We'll be back with updates as more 2026 data rolls in.