Dynasty & keeper values
NFL dynasty values
What each player is worth over the next three seasons: his production rate, how players his age at his position have aged in our logs, and how many games players like him actually play — on a 0–100 scale above replacement.
5
Josh AllenQB1BUFage 30.472.8T2 · Elite
Rate 22.4 PPR ppg = production 23 blended with opportunity-implied 19. Sample: 2026: 5g 28.6, 2025: 16g 22.8, 2024: 16g 23.7 (37 games). Draft: 2018 R1.7. Expected weighted points: 638.
16
Patrick MahomesQB2KCage 31.149.8T3 · Core starter
Rate 18.8 PPR ppg = production 19 blended with opportunity-implied 17.6. Sample: 2026: 4g 20.9, 2025: 14g 20.4, 2024: 16g 17.7 (34 games). Draft: 2017 R1.10. Expected weighted points: 546.
Guests see the top 25 overall. Members get the full board (258 players), every scoring format, 2027 rookie pick values and the trade calculator with up to 8 assets per side.
How good is this?
We rebuilt the values as of the start of past seasons using only data available then (curves, weights and draft priors refitted on earlier seasons) and compared the order with what players actually scored over the next three seasons (PPR, weighted 1.0 / 0.8 / 0.6). Rank correlation (Spearman): 1 = perfect order, 0 = random. Baseline = ranking by last season's total points.
| As of | Our top 200 | Baseline's top 200 | Top-24 hits |
|---|---|---|---|
| 2022 (2022-2024) | 0.55 vs 0.50 | 0.69 vs 0.60 | 10/24 vs 12 |
| 2023 (2023-2025) | 0.61 vs 0.41 | 0.61 vs 0.53 | 11/24 vs 9 |
Read: ours vs baseline on the same players. By position (our top 40 QB / 60 RB / 80 WR / 40 TE): 2022 QB 0.53 vs 0.48, RB 0.64 vs 0.36, WR 0.61 vs 0.60, TE 0.61 vs 0.53; 2023 QB 0.61 vs 0.49, RB 0.65 vs 0.25, WR 0.67 vs 0.47, TE 0.41 vs 0.50. Incoming rookies (valued from draft slot only): 2022 0.52 (n=79), 2023 0.69 (n=82). Summary: Overall better than the baseline (avg +0.08); clearly better at QB/RB/WR; about even at TE; top-24 hits about even (avg +0.0). 2 backtest seasons is a small sample — treat values as a well-reasoned starting point, not a guarantee.
How the values are built
Inputs. Our NFL game logs (nflverse weekly stats, regular season, 2015-2026 (through week 5)); birth dates and NFL draft slot from the nflverse player file. A “game” is a week with a stat line. QB/RB/WR/TE only.
1. Production rate. Points per game over his last three seasons weighted 1 / 0.35 / 0.1 (newest first), plus 4 pseudo-games at the average for his position and draft round in years 1–3 (so a 3-game sample can't look like a star). Blended 15% with an opportunity-implied rate — points per game predicted from target share (WR/TE), carry share + target share (RB) or pass attempts + carries (QB); R² QB 0.69, RB 0.9, WR 0.88, TE 0.86. The weights, prior size and blend were chosen to best predict next-season points per game on past seasons (error ±3.23 ppg).
2. Age curve. For every player with 6+ games in back-to-back seasons we measured the change in points per game by age, per position (2015–2025), and smoothed it with a Gaussian kernel. Examples (RB): 22: +0.66, 24: +0, 26: -0.48, 28: -0.83, 30: -1.08 ppg per year. Only survivors are in this curve, so it understates decline; retirement and lost jobs live in step 3.
3. Expected games. How many games players at his position, age band and quality level actually logged 1, 2 and 3 seasons later in our data — zero if they vanished. For 2026 we count only his team's remaining games (week 5 done).
4. Value. Expected points per season, weighted 1.0 / 0.8 / 0.6 across 2026, 2027, 2028, counted only above replacement (12-team 1QB: QB13, RB30, WR42, TE13 in that season), then scaled so #1 = 100. Tiers: Tier 1 ≥ 80 · 2 ≥ 60 · 3 ≥ 40 · 4 ≥ 25 · 5 ≥ 12 · 6 > 0. Superflex, TE premium and longer horizons aren't modelled yet.
Not in the number: injuries in progress, contracts, depth-chart changes, coaching. Rookies with no NFL games are valued from their draft slot only. Updated 2026-10-16.