How accurate are our projections?
Most fantasy services tell you their projections are good. We show you.
Below is how our current models score on this season's games so far: every graded game measured against what actually happened, side-by-side with what FantasyPros, ESPN, and Sleeper said for the same players. If you sign up today, these are the projections you get.
2026 season to date, Weeks 1–5
Our current models, re-run week by week using only information available before each game, then graded against what actually happened. Each game is graded on the projections those models would have had on the board when it kicked off.
Every week below was scored by one version of our models, as of October 15, 2026.
What a re-run cannot rebuild as it stood: the list of players and their clubs is today's, and any stat corrections made since a week was played are included.
Not in this record yet: Week 6, which has been played but not yet re-run with this version of our models. We add a week only once it has been.
| Stat | Outcomes | 50% range | 80% range | 90% range | Our error | vs ESPN | vs Sleeper | vs FantasyPros |
|---|---|---|---|---|---|---|---|---|
| Receptions | 1681 | 44% (41–46) | 74% (72–75) | 85% (84–87) | 1.30 | +0.07 (+0.04 to +0.10) · worse · n=1435 | +0.10 (+0.07 to +0.14) · worse · n=1502 | +0.05 (-0.05 to +0.16) · level · n=153 |
| Fumbles lost | 1597 | 49% (48–49) | 78% (77–79) | 88% (87–89) | 0.09 | +0.00 (-0.00 to +0.01) · level · n=1399 | +0.01 (+0.01 to +0.01) · worse · n=1095 | -0.00 (-0.02 to +0.01) · level · n=107 |
| Receiving first downs | 1586 | 47% (45–48) | 75% (74–77) | 85% (84–86) | 0.90 | — | — | — |
| Receiving TDs | 1558 | 49% (48–50) | 78% (77–79) | 87% (86–88) | 0.26 | +0.02 (+0.02 to +0.03) · worse · n=1340 | +0.02 (+0.01 to +0.03) · worse · n=1351 | -0.01 (-0.02 to +0.00) · level · n=152 |
| Receiving yards | 1553 | 48% (45–50) | 81% (79–83) | 90% (89–92) | 17.5 | +0.9 (+0.5 to +1.3) · worse · n=1347 | +1.3 (+0.9 to +1.8) · worse · n=1404 | -0.1 (-1.5 to +1.2) · level · n=153 |
| Rushing TDs | 612 | 48% (47–50) | 77% (76–79) | 87% (86–89) | 0.32 | -0.00 (-0.01 to +0.01) · level · n=524 | -0.00 (-0.01 to +0.01) · level · n=535 | -0.01 (-0.03 to +0.01) · level · n=104 |
| Rushing first downs | 595 | 50% (46–53) | 78% (76–81) | 88% (86–90) | 1.15 | — | — | — |
| Rushing yards | 544 | 44% (40–48) | 81% (77–84) | 91% (89–94) | 17.4 | -0.1 (-0.9 to +0.7) · level · n=488 | +0.3 (-0.4 to +1.1) · level · n=508 | +0.9 (-0.2 to +2.0) · level · n=103 |
| Passing first downs | 178 | 53% (45–61) | 81% (75–87) | 91% (87–95) | 3.32 | — | — | — |
| Passing yards | 178 | 52% (43–60) | 81% (75–86) | 90% (85–94) | 63.1 | +0.1 (-3.0 to +3.1) · level · n=160 | -3.9 (-7.5 to -0.4) · better · n=155 | -0.5 (-4.4 to +3.3) · level · n=54 |
| Interceptions | 177 | 46% (43–50) | 76% (72–80) | 87% (84–90) | 0.64 | -0.01 (-0.04 to +0.02) · level · n=159 | +0.00 (-0.04 to +0.04) · level · n=155 | -0.00 (-0.04 to +0.03) · level · n=54 |
| Passing TDs | 174 | 47% (41–54) | 74% (69–79) | 86% (82–89) | 0.88 | +0.04 (-0.01 to +0.08) · level · n=157 | +0.03 (-0.02 to +0.08) · level · n=155 | +0.03 (-0.03 to +0.09) · level · n=54 |
| Team defense — blocked kicks | 158 | 42% (39–45) | 72% (69–75) | 83% (79–86) | 0.13 | — | — | — |
| Team defense — fumble rec. | 158 | 44% (42–46) | 72% (69–75) | 81% (78–85) | 0.66 | — | — | — |
| Team defense — INTs | 158 | 45% (40–49) | 74% (70–77) | 86% (84–88) | 0.68 | — | — | — |
| Team defense — pts allowed | 158 | 47% (39–55) | 77% (69–85) | 91% (86–95) | 7.4 | — | — | — |
| Team defense — sacks | 158 | 61% (53–70) | 80% (72–88) | 87% (82–93) | 1.20 | — | — | — |
| Team defense — safeties | 158 | 53% (50–56) | 83% (79–88) | 91% (88–94) | 0.02 | — | — | — |
| Team defense — TDs | 158 | 50% (48–52) | 80% (77–83) | 89% (87–92) | 0.19 | — | — | — |
| FGs made (≤19 yd) | 147 | 55% (50–60) | 87% (84–89) | 93% (92–95) | 0.02 | — | — | — |
| FGs made (20-29 yd) | 147 | 49% (47–51) | 79% (77–81) | 89% (87–91) | 0.51 | — | +0.04 (+0.02 to +0.06) · worse · n=146 | — |
| FGs made (30-39 yd) | 147 | 47% (45–50) | 76% (72–80) | 86% (83–89) | 0.63 | — | +0.03 (+0.00 to +0.05) · worse · n=146 | — |
| FGs made (40-49 yd) | 147 | 46% (43–49) | 74% (70–78) | 83% (79–87) | 0.61 | — | +0.02 (+0.00 to +0.03) · worse · n=146 | — |
| FGs made (50-59 yd) | 147 | 47% (43–52) | 75% (71–80) | 85% (80–90) | 0.55 | — | — | — |
| FGs made (50+ yd) | 147 | 46% (42–50) | 75% (71–80) | 85% (80–90) | 0.57 | — | +0.05 (+0.01 to +0.09) · worse · n=89 | — |
| FGs missed (20-29 yd) | 147 | 56% (54–58) | 87% (84–90) | 92% (89–96) | 0.02 | — | — | — |
| FGs missed (30-39 yd) | 147 | 59% (56–62) | 83% (81–85) | 91% (89–93) | 0.05 | — | — | — |
| FGs missed (40-49 yd) | 147 | 44% (41–47) | 70% (65–75) | 85% (82–89) | 0.23 | — | — | — |
| FGs missed (50-59 yd) | 147 | 46% (43–50) | 74% (68–80) | 86% (82–91) | 0.28 | — | — | — |
| Extra points made | 147 | 54% (44–63) | 80% (73–88) | 85% (78–91) | 1.07 | — | +0.02 (-0.03 to +0.06) · level · n=146 | +0.04 (-0.03 to +0.10) · level · n=50 |
| FGs missed (60+ yd) | 120 | 50% (46–54) | 80% (74–86) | 86% (80–92) | 0.07 | — | — | — |
| FGs made (60+ yd) | 80 | — | — | — | 0.06 | — | — | — |
| FGs missed (≤19 yd) | 40 | — | — | — | 0.00 | — | — | — |
| Individual defenders (DL, LB, DB) | ||||||||
| Assisted tackles | 2460 | 54% (52–56) | 73% (71–75) | 83% (82–84) | 1.08 | — | — | — |
| Solo tackles | 2460 | 50% (48–52) | 73% (71–74) | 83% (82–84) | 1.17 | — | — | — |
| Fumble recoveries | 2454 | 61% (60–62) | 83% (83–84) | 90% (89–91) | 0.06 | — | — | — |
| Forced fumbles | 2452 | 52% (51–53) | 77% (76–78) | 85% (84–86) | 0.10 | — | — | — |
| Tackles for loss | 2450 | 48% (47–49) | 77% (76–78) | 87% (86–88) | 0.29 | — | — | — |
| Passes defended | 2277 | 47% (46–48) | 75% (74–76) | 85% (84–86) | 0.38 | — | — | — |
| Defensive TDs | 2192 | 64% (63–64) | 87% (87–87) | 94% (94–94) | 0.01 | — | — | — |
| Sacks | 1789 | 48% (47–49) | 75% (74–77) | 86% (85–87) | 0.27 | — | — | — |
| QB hits | 1736 | 48% (47–50) | 77% (75–78) | 87% (85–88) | 0.53 | — | — | — |
| Interceptions (defense) | 1526 | 41% (39–42) | 72% (71–74) | 83% (81–84) | 0.13 | — | — | — |
A range column is how often the outcome landed inside the range we published at that width, with a 95% interval in brackets. "Our error" is the average miss of the number we published. A "vs" column is our average miss minus that source's, on the players both of us projected: positive means we missed by more. "Level" means the 95% interval includes zero. Intervals treat each player's weeks and stats as one piece of evidence, so a player counted many times does not make a gap look more certain than it is. Stats with fewer than 5 outcomes, and pairings with fewer than 5 shared outcomes, are left out, and so is any figure whose outcomes all come from one player, which leaves no interval to state.
Some sources store a zero where they mean "no projection for this player", which is not the same claim as projecting zero. We drop any value too small to be a projection on that stat's own scale, and here that withheld 337 from Sleeper and 102 from FantasyPros. Those outcomes are left out of both sides of that pairing, ours included.
A game whose outcome was exactly zero counts toward a range only for the share of the chance of a zero we projected that falls inside it, instead of counting as inside every range that starts at zero. On stats that are usually zero, the old count made even a narrow range look as if it almost never missed. This changed the printed range figures for every stat in this table.
Left out of the table: outcomes where we projected close to nothing for that stat (under 10% of the stat's 90th-percentile projection), 5914 in all.
Individual defenders are graded stat by stat, in the stat's own units, against each game's defensive box score. No outside source publishes projections for individual defenders, so those rows have nothing to compare against. 124 defender games could not be matched to a player we track and are left out.
Fantasy points, starters and backups
The same graded games, added up into fantasy points for each player's week, and our average miss compared with each source's on the players both of us projected. A cell reads our gap first (positive means we missed by more), then our average miss against theirs.
| Position | Group | Player-weeks | vs ESPN | vs Sleeper | vs FantasyPros |
|---|---|---|---|---|---|
| QB | Starters | 146 | +0.16 (-0.08 to +0.39) · level · 6.28 vs 6.12 · n=146 | +0.01 (-0.34 to +0.36) · level · 6.20 vs 6.19 · n=145 | +0.17 (-0.19 to +0.54) · level · 7.07 vs 6.90 · n=54 |
| QB | Backups | 37 | +0.39 (-1.27 to +2.05) · level · 8.20 vs 7.81 · n=14 | -0.74 (-3.07 to +1.58) · level · 7.93 vs 8.68 · n=10 | — |
| RB | Starters | 292 | -0.21 (-0.44 to +0.03) · level · 5.14 vs 5.35 · n=286 | -0.05 (-0.25 to +0.16) · level · 5.13 vs 5.18 · n=287 | +0.20 (-0.20 to +0.60) · level · 7.02 vs 6.82 · n=52 |
| RB | Backups | 160 | +0.59 (+0.09 to +1.08) · worse · 3.18 vs 2.60 · n=85 | +0.47 (-0.02 to +0.95) · level · 3.11 vs 2.64 · n=102 | — |
| WR | Starters | 429 | -0.05 (-0.20 to +0.10) · level · 5.57 vs 5.62 · n=429 | +0.01 (-0.15 to +0.17) · level · 5.59 vs 5.58 · n=425 | +0.24 (-0.33 to +0.82) · level · 8.88 vs 8.63 · n=52 |
| WR | Backups | 304 | +0.52 (+0.27 to +0.78) · worse · 3.46 vs 2.94 · n=209 | +0.74 (+0.51 to +0.97) · worse · 3.34 vs 2.60 · n=242 | — |
| TE | Starters | 149 | +0.10 (-0.10 to +0.30) · level · 5.55 vs 5.45 · n=149 | +0.01 (-0.21 to +0.23) · level · 5.58 vs 5.56 · n=148 | +0.01 (-0.49 to +0.51) · level · 6.86 vs 6.86 · n=49 |
| TE | Backups | 352 | +0.69 (+0.47 to +0.91) · worse · 3.16 vs 2.47 · n=264 | +0.85 (+0.64 to +1.06) · worse · 3.09 vs 2.23 · n=305 | — |
| All four | Starters | 1016 | -0.04 (-0.14 to +0.06) · level · 5.55 vs 5.59 · n=1010 | -0.01 (-0.11 to +0.10) · level · 5.55 vs 5.55 · n=1005 | +0.16 (-0.07 to +0.39) · level · 7.46 vs 7.30 · n=207 |
| All four | Backups | 853 | +0.61 (+0.45 to +0.77) · worse · 3.40 vs 2.79 · n=572 | +0.73 (+0.58 to +0.88) · worse · 3.26 vs 2.53 · n=659 | — |
Fantasy points are full PPR over nine stats every source publishes, in points per unit: passing yards 0.04, passing TDs 4, interceptions −2, rushing yards 0.1, rushing TDs 6, receptions 1, receiving yards 0.1, receiving TDs 6, fumbles lost −2. A starter is a player in a team's QB1, RB1–2, WR1–3, TE1 slot on that week's depth chart; every other graded player at those positions is a backup, including players the chart does not list. The split reads no projection, ours or anyone else's. A source counts as projecting a player-week only when it published passing yards for a quarterback, rushing yards for a running back, receiving yards for a wide receiver or tight end; each comparison is over exactly those player-weeks, ours measured on the same ones. "All four" is every QB, RB, WR and TE of that kind pooled into one population, not an average of the rows above it. "Level" means the 95% interval includes zero, and intervals treat each player's weeks as one piece of evidence. A cell with fewer than 5 shared player-weeks is left out.
Player-weeks no source projected are in no comparison here: 5 starter player-weeks and 144 backup player-weeks.
How our longer-range projections tested
Before we publish a projection for the rest of the season or a season ahead, we test it on past seasons: projections built only from what was known at the time, checked against what happened. These are those tests, for what we publish today. Starters and non-starters are tested separately and never added together.
S is the season a projection was made in. Every figure here is graded on past seasons whose results are known, never on seasons still to be played.
The tests behind these tables recorded the 80% and 90% ranges only, so the 50% column reads “not measured” rather than a figure we did not take.
We do not offer five-season or rest-of-career totals, so there is nothing of that kind to grade.
Rest of season
Ranges and the average miss are in fantasy points for the whole period, scored the way our dynasty and season pages score it.
Starters (QB, DL, LB, DB): measured at weeks 4, 8 and 12 of the seasons 2016 to 2024, on the rest of each season; graded October 2, 2026. Non-starters (QB, RB, WR, TE, DL, LB, DB): measured at weeks 4, 8 and 12 of the seasons 2016 to 2024, on the rest of each season; graded October 2, 2026. Starters (RB, WR): measured at weeks 4, 8 and 12 of the seasons 2016 to 2024, on the rest of each season; graded October 3, 2026. Starters (TE): measured at weeks 4, 8 and 12 of the seasons 2016 to 2024, on the rest of each season; graded October 5, 2026.
Next season (S+1)
Ranges and the average miss are in fantasy points for the whole period, scored the way our dynasty and season pages score it.
Starters (QB, LB): measured at weeks 4, 8 and 12 of the seasons 2016 to 2024, on the following season; graded October 16, 2026. Non-starters (QB, RB, TE, DL, LB, DB): carried forward from the seasons 2017 to 2023, measured two seasons later; graded October 16, 2026. Starters (RB, TE): measured at weeks 4, 8 and 12 of the seasons 2017 to 2022, on the following season; graded October 16, 2026. Starters (WR): measured at weeks 4, 8 and 12 of the seasons 2017 to 2022, on the following season; a player starting this season after a season below the starter mark is counted as a starter; graded October 18, 2026. Non-starters (WR): carried forward from the seasons 2017 to 2023, measured two seasons later; a player starting this season after a season below the starter mark is counted as a starter; graded October 18, 2026. Starters (DL): measured at weeks 4, 8 and 12 of the seasons 2018 to 2024, on the following season, with the range narrowed as served; graded October 22, 2026. Starters (DB): measured at weeks 4, 8 and 12 of the seasons 2016 to 2024, on the following season; graded October 15, 2026.
Two seasons ahead (S+2)
Ranges and the average miss are in fantasy points for the whole period, scored the way our dynasty and season pages score it.
Starters (QB): measured at weeks 4, 8 and 12 of the seasons 2016 to 2023, two seasons on; graded October 16, 2026. Non-starters (QB, RB, TE, DL, LB, DB): carried forward from the seasons 2017 to 2021; graded October 15, 2026. Starters (RB): measured at weeks 4, 8 and 12 of the seasons from 2017 on, against seasons through 2025; graded October 16, 2026. Starters (WR): measured at weeks 4, 8 and 12 of the seasons from 2017 on, against seasons through 2025; a player starting this season after a season below the starter mark is counted as a starter; graded October 18, 2026. Non-starters (WR): carried forward from the seasons 2017 to 2021; a player starting this season after a season below the starter mark is counted as a starter; graded October 18, 2026. Starters (TE, DL, LB, DB): measured at weeks 4, 8 and 12 of the seasons from 2017 on, against seasons through 2025; graded October 15, 2026.
A few starting defenders whose newest full season is an earlier one are served for this season by a construction we have not graded, because our history holds too few such players to test it; the figures above are for the rest.
Three seasons ahead (S+3)
Ranges and the average miss are in fantasy points for the whole period, scored the way our dynasty and season pages score it.
Starters (QB): measured at weeks 4, 8 and 12 of the seasons from 2017 on, against seasons through 2025; graded October 23, 2026. Non-starters (QB, TE): carried forward from the seasons 2017 to 2021; graded October 16, 2026. Starters (RB): measured at weeks 4, 8 and 12 of the seasons from 2017 on, against seasons through 2025; graded October 16, 2026. Non-starters (RB, DL, LB, DB): carried forward from the seasons 2017 to 2021; graded October 15, 2026. Starters (WR): measured at weeks 4, 8 and 12 of the seasons from 2017 on, against seasons through 2025; a player starting this season after a season below the starter mark is counted as a starter; graded October 18, 2026. Non-starters (WR): carried forward from the seasons 2017 to 2021; a player starting this season after a season below the starter mark is counted as a starter; graded October 18, 2026. Starters (TE, DL, LB): measured at weeks 4, 8 and 12 of the seasons from 2017 on, against seasons through 2025; graded October 15, 2026.
Rest of season plus S+1
Ranges and the average miss are in fantasy points for the whole period, scored the way our dynasty and season pages score it.
Starters (QB): measured at weeks 4, 8 and 12 of the seasons 2020 to 2024; graded October 6, 2026. Non-starters (QB, RB, TE, DL, LB, DB): measured at weeks 4, 8 and 12 of the seasons 2016 to 2024; graded October 5, 2026. Starters (RB, TE): measured at weeks 4, 8 and 12 of the seasons 2019 to 2024; graded October 6, 2026. Starters (WR): measured at weeks 4, 8 and 12 of the seasons 2021 to 2024; graded October 17, 2026. Non-starters (WR): measured at weeks 4, 8 and 12 of the seasons 2016 to 2024; a player starting this season after a season below the starter mark is counted as a starter; graded October 18, 2026. Starters (DL): measured at weeks 4, 8 and 12 of the seasons 2018 to 2024; graded October 22, 2026. Starters (LB, DB): measured at weeks 4, 8 and 12 of the seasons 2016 to 2024; graded October 21, 2026.
Rest of season through S+2
Ranges and the average miss are in fantasy points for the whole period, scored the way our dynasty and season pages score it.
Starters (RB, WR): measured at weeks 4, 8 and 12 of the seasons 2020 to 2023; graded October 16, 2026. Non-starters (RB, TE): measured at weeks 4, 8 and 12 of the seasons 2016 to 2023; graded October 5, 2026. Non-starters (WR): measured at weeks 4, 8 and 12 of the seasons 2016 to 2023; a player starting this season after a season below the starter mark is counted as a starter; graded October 18, 2026. Starters (TE): measured at weeks 4, 8 and 12 of the seasons 2020 to 2023; graded October 6, 2026. Starters (DL): measured at weeks 4, 8 and 12 of the seasons 2018 to 2023; graded October 22, 2026. Non-starters (DL, LB, DB): measured at weeks 4, 8 and 12 of the seasons 2016 to 2023; graded October 22, 2026. Starters (LB, DB): measured at weeks 4, 8 and 12 of the seasons 2017 to 2023; graded October 21, 2026.
Rest of season through S+3
Ranges and the average miss are in fantasy points for the whole period, scored the way our dynasty and season pages score it.
Non-starters (RB, WR, TE, DL, LB): measured at weeks 4, 8 and 12 of the seasons 2018 to 2022; graded October 24, 2026.
S+1 through S+3 combined
Ranges and the average miss are in fantasy points for the whole period, scored the way our dynasty and season pages score it.
Starters (QB): measured at weeks 4, 8 and 12 of the seasons 2021 to 2022; graded October 23, 2026. Non-starters (QB, TE): carried forward from the seasons 2017 to 2021 during the next season, measured on the three seasons after it; graded October 16, 2026. Starters (RB): measured at weeks 4, 8 and 12 of the seasons 2020 to 2022; graded October 16, 2026. Non-starters (RB): carried forward from the seasons 2017 to 2021 during the next season, measured on the three seasons after it; graded October 21, 2026. Starters (WR): measured at weeks 4, 8 and 12 of the seasons 2019 to 2022; a player starting this season after a season below the starter mark is counted as a starter; graded October 18, 2026. Non-starters (WR): carried forward from the seasons 2017 to 2021 during the next season, measured on the three seasons after it; a player starting this season after a season below the starter mark is counted as a starter; graded October 18, 2026. Starters (TE): measured at weeks 4, 8 and 12 of the seasons 2019 to 2022; graded October 16, 2026. Starters (DL, LB): measured at weeks 4, 8 and 12 of the seasons 2018 to 2022, on the three seasons after it; graded October 20, 2026. Non-starters (DL, LB, DB): carried forward from the seasons 2017 to 2021 during the next season, measured on the three seasons after it; graded October 22, 2026.
A range column is how often the outcome landed inside the range we would have published at that width. "Rows tested" counts player-seasons, or player-weeks where a test was taken at several points in a season, so one player can count more than once. "Average miss" is how far the middle of our projection landed from what happened. Each table states when its tests were taken; we do not combine tests taken at different times into one figure.
How we measure
We score the per-stat projections our models publish, the numbers and the ranges, with nothing chosen after the fact. Every player who played and had a projection from us is included for the stats he recorded. A player's fantasy-point floor and ceiling are built from these ranges and then adjusted for whether he was expected to play; that adjustment is not scored here.
We rely on calibrated uncertainty: projections that come with honest prediction intervals, not just point estimates. As more weeks land, this page adds them, and you can see whether the calibration holds up across the season. If it doesn't, you'll see that here too. When we change our models, we re-run the season with the new version before this page describes it.
Where we get our college numbers
A rookie has no NFL record from which to project, so his first-season numbers lean on what he did in college. Our college numbers come from CollegeFootballData.com, a freely accessible, independently run college-football data project worth knowing about if you follow the draft.