Projection model

Fantasy Football Model

A 1-point-PPR projection for every position, trained on 27 seasons and measured against the only benchmark that matters.

Player-weeks trained on: 153,026; 1999–2025
Seasons back-tested: 20; each fitted only on what came before
Best gain over recent form: 15%; defense, where the market carries it
Worst gain over recent form: 7%; the honest number for a noisy target

2026, before it happens

Every projection here rests on a record that stops at the end of 2025. Two limits are worth knowing before the board is used as a draft aid.

No rookies. A player with no prior NFL game has no history to project from, so he is absent rather than guessed at. This is a board of returning players.
Most of the schedule is not priced yet. The betting market has posted lines for part of the season. Where it has not, the input that carries kickers and defenses is missing, so those two boards are the softest here, and the accuracy tab shows why.

RBRunning back

Top 40 of 97 eligible

#	Player	Team	Season	Per game	Floor	Ceiling	Prior games
1	Bijan Robinson	ATL	343	20.1	6.4	31.9	50
2	Christian McCaffrey	SF	335	19.7	7.7	31.3	111
3	De'Von Achane	MIA	290	17.0	4.7	26.8	43
4	Jahmyr Gibbs	DET	287	16.9	6.8	27.0	48
5	Ashton Jeanty	LV	274	16.1	7.0	26.4	16
6	Derrick Henry	BAL	269	15.8	6.0	23.9	151
7	Chase Brown	CIN	264	15.5	6.6	25.6	44
8	Jonathan Taylor	IND	251	14.7	5.4	25.0	83
9	Kyren Williams	LAR	234	13.7	4.6	24.1	52
10	Rhamondre Stevenson	NE	221	13.0	4.0	24.4	69
11	Tyrone Tracy Jr.	NYG	214	12.6	3.7	22.8	31
12	Jaylen Warren	PIT	211	12.4	4.0	22.2	63
13	Kenneth Gainwell	TB	210	12.3	3.3	21.3	82
14	Saquon Barkley	PHI	207	12.2	6.1	22.0	105
15	Travis Etienne	NO	203	11.9	4.3	22.3	65
16	D'Andre Swift	CHI	201	11.8	3.6	21.9	88
17	RJ Harvey	DEN	200	11.8	3.6	21.5	16
18	Kenneth Walker III	KC	196	11.6	2.9	22.4	57
19	Zach Charbonnet	SEA	196	11.5	3.4	22.5	48
20	James Cook	BUF	195	11.5	4.5	23.4	65
21	Bucky Irving	TB	190	11.2	3.6	20.3	26
22	Aaron Jones	MIN	188	11.1	3.4	19.5	123
23	Tony Pollard	TEN	188	11.0	3.5	20.8	110
24	Breece Hall	NYJ	183	10.7	3.1	20.1	55
25	J.K. Dobbins	DEN	182	10.7	2.1	20.4	46
26	Rico Dowdle	PIT	178	10.4	2.1	19.4	50
27	Audric Estimé	NO	177	10.4	2.2	20.0	17
28	TreVeyon Henderson	NE	158	9.3	2.1	19.7	16
29	Javonte Williams	DAL	157	9.3	3.6	21.6	69
30	Josh Jacobs	GB	157	9.3	3.0	20.1	104
31	James Conner	ARI	152	8.9	2.1	18.7	107
32	Kyle Monangai	CHI	145	8.5	1.3	18.1	16
33	Patrick Taylor	SF	145	8.5	1.2	17.8	25
34	Chris Rodriguez Jr.	JAX	142	8.4	1.6	18.1	27
35	Michael Carter	TEN	142	8.4	1.2	17.6	60
36	Isiah Pacheco	DET	142	8.3	1.2	17.4	50
37	Jacory Croskey-Merritt	WAS	137	8.1	1.3	16.4	16
38	Tyjae Spears	TEN	137	8.0	1.6	17.5	41
39	Bam Knight	ARI	131	7.7	0.6	15.4	20
40	Blake Corum	LAR	125	7.3	0.8	16.7	31

Floor and ceiling are per game, not per season: the tenth and ninetieth percentile of a single week. Adding a floor across seventeen weeks would describe a year in which the same player is unlucky every Sunday, which is not a floor anybody should plan around.

Every player on this board has at least 16 prior games, a full season. Held off it for a shorter record: Omarion Hampton (8 games, would have ranked 10); Cam Skattebo (7 games, would have ranked 14); Quinshon Judkins (13 games, would have ranked 30); Devin Neal (8 games, would have ranked 34); Trey Benson (15 games, would have ranked 36); Sincere McCormick (3 games, would have ranked 39). The projections themselves are unchanged. The model still carries every one of them, and the full board is in the model’s own CSV.