Season-by-season production from Pro Football Reference, with fantasy points computed under Superflex PPR (1 PPR · 4 pt pass TD · 6 pt rush/rec TD · −2 INT · −2 fumble). Career totals on the bottom row.
| Year | Team | GP | Rush Att | Rush Yds | Rush TD | Tgt | Rec | Rec Yds | Rec TD | FP |
|---|---|---|---|---|---|---|---|---|---|---|
| 2020 | BAL | 15 | 134 | 805 | 9 | 24 | 18 | 120 | 0 | 160.5 |
| 2022 | BAL | 8 | 92 | 520 | 2 | 8 | 7 | 42 | 1 | 81.2 |
| 2023 | BAL | 1 | 8 | 22 | 1 | 3 | 2 | 15 | 0 | 11.7 |
| 2024 | LAC | 13 | 195 | 905 | 9 | 38 | 32 | 153 | 0 | 191.8 |
| 2025 | DEN | 10 | 153 | 772 | 4 | 14 | 11 | 37 | 0 | 115.9 |
| Career | — | 47 | 582 | 3024 | 25 | 87 | 70 | 367 | 1 | 561.1 |
The top-10 most similar long-arc NFL players matched by fantasy-point production curve at this career stage. Each comp's "Peak 3yr fp/g" is their best 3-season fp/g average under Superflex PPR (era-pace-adjusted to modern). Similarity is bounded in (0, 1] — 1.0 means an identical career-stage profile vector. "Career" notes the comp's NFL longevity; the washed out badge flags comps whose career ended by age 30 with fewer than 8 NFL seasons (the engine’s bust definition). The ⏳ 1985 badge marks comps whose snapshot season predates 1999 — the corpus now spans 1980-2025, so Payton, Emmitt, Marcus Allen, Rice, Marino-era greats are eligible matches. Pre-1999 comps carry a 0.9× confidence haircut because era-pace adjustment is principled but not perfect.
| Comparable | Pos | Last season | Similarity | Peak 3yr fp/g | Career | Post-age seasons | Career fp | Projected pts |
|---|---|---|---|---|---|---|---|---|
| James White washed out | RB | 2021 | 0.777 | 12.2 | 6 seasons · ended age 28 | 1 | 918 | 117 |
| Carlos Hyde | RB | 2021 | 0.742 | 13.7 | 8 seasons · ended age 31 | 4 | 962 | 369 |
| Michael Bush washed out | RB | 2014 | 0.741 | 10.7 | 6 seasons · ended age 29 | 2 | 707 | 138 |
| Pierre Thomas | RB | 2015 | 0.734 | 13.6 | 9 seasons · ended age 31 | 4 | 1214 | 459 |
| Ryan Mathews washed out | RB | 2016 | 0.725 | 13.8 | 7 seasons · ended age 29 | 2 | 1060 | 264 |
| Willis McGahee | RB | 2013 | 0.721 | 13.1 | 10 seasons · ended age 32 | 5 | 1566 | 570 |
| Latavius Murray | RB | 2023 | 0.720 | 12.1 | 10 seasons · ended age 33 | 6 | 1391 | 667 |
| Lorenzo White ⏳ 1993 washed out | RB | 1995 | 0.697 | 12.0 | 7 seasons · ended age 29 | 2 | 859 | 152 |
| Raymont Harris ⏳ 1997 washed out | RB | 2000 | 0.695 | 11.0 | 4 seasons · ended age 28 | 1 | 457 | 39 |
| Duce Staley | RB | 2006 | 0.690 | 13.6 | 8 seasons · ended age 30 | 3 | 1132 | 251 |
Every component below is sourced from the engine output
(engine_rankings.json) and the displayed comp table. The
final production score is 211.
fantasy_arc_v2
Projection path: peak_anchored
Comps: 25
Avg similarity (top 20): 0.698
Comp pool washed-out rate: 55% (11/20)
| Step | Value | What it is |
|---|---|---|
| Comp-weighted projection (v3.3 primary) | 316 | Σ (simi × post-age-fpi) / Σ simi. Sanity check from the top-20 row data on this page: 307. This is the “weighted average of the comparable players’ actual realised post-age fantasy points” — Phil's 2026-05-28 brief. |
| Peak-anchored projection | 651 | The player's own peak-3yr-fp/g × expected games × horizon × 5%/yr time discount, capped at 1.25× the top single-comp projection. Used as a soft blend for elite-tier producers (60/40 comp-weighted / capped peak) so a Mahomes / Allen isn't fully dragged to the comp-pool mean. |
| Banked-credit floor (diagnostic only) | 890 | The v3.1 “dynasty equity” floor = career_total_fp × yrs-rem-weight + recent rate × short forward window. v3.3 no longer lets this override the projection — banked production is already realised, not “remaining.” Shown for transparency. |
| Raw projection (pre-penalty) | 452 | Blended comp-weighted + capped peak-anchored, per the v3.3 methodology described above. |
| × Survival | 0.550 | Multiplier reflecting how many comps washed out by age 30. Today: 1 − 0.5 × bust_rate. |
| × Sample confidence | 1.000 | Shrinks small-sample projections toward the position-tier baseline when raw > baseline; multiplies straight otherwise. = max(249 × confidence + baseline × (1−conf), or 249 × 1.000) when below baseline. |
| × Late-breakout penalty | 1.000 | QB-only. Scales by confidence so unproven late-breakout rookies only pay a fraction of the discount. |
| × Missed-season penalty (v3.3) | 0.847 | Multiplicative haircut for players who didn't play (or only played a partial slate) in the most recent NFL season. 0.70 for a full missed season, 0.85 for <8 games, 0.45 for two+ full missed seasons. only 10 of 17 games in 2025 (partial season — v3.7 scaled penalty) |
| = Final production score | 211 | Drives the player's rank on the Similarity Scores page. |
Want this player ranked under your league's specific scoring + roster rules? Head to Dynasty Rankings.