Through Week 5: How DFS Projections Are Holding Up in 2026
0Through Week 5 of 2026, the average projection error across six sources was higher at every position than in Weeks 1-5 of 2024 and 2025. Results from Week 6 onward are not included. TE showed the largest increase: an average miss of 6.46 DraftKings points per player-week, compared with 5.14 in 2024 and 5.02 in 2025.
Differences between providers were much smaller with less than half a point of average error separating the best and worst source at every position. The FFA Average ranked in the top three at all four positions, but five weeks offers little basis for declaring a decisive winner.
Several players illustrate how large individual misses have been. The FFA Average projected Jaxon Smith-Njigba for 107.8 points over five weeks, he scored 153.9. It projected De’Von Achane for 52.1 over three weeks, he scored 24.6.
How Sources Have Done Through Week 5
Each source is scored on the same players: FFA Average’s top 20 QB, 40 RB, 40 WR and 20 TE each week, where all six sources had a projection for the player. The measure is mean absolute error (MAE), the average number of DK points a projection missed by per player-week. Lower is better.
| Source | QB | RB | WR | TE |
|---|---|---|---|---|
| CBS | 6.35 (2) | 6.39 (4) | 7.91 (5) | 6.45 (5) |
| ESPN | 6.31 (1) | 6.43 (5) | 7.61 (1) | 6.44 (4) |
| FFA Average | 6.38 (3) | 6.27 (2) | 7.67 (2) | 6.35 (1) |
| FFToday | 6.65 (6) | 6.25 (1) | 7.98 (6) | 6.71 (6) |
| FantasySharks | 6.47 (4) | 6.67 (6) | 7.83 (4) | 6.39 (2) |
| numberFire | 6.53 (5) | 6.38 (3) | 7.77 (3) | 6.41 (3) |
| App DK | 6.39 | 6.26 | 7.67 | 6.35 |
| Six-source mean | 6.45 | 6.40 | 7.79 | 6.46 |
Ranks among the six sources are in brackets. App DK is the FFA app’s DraftKings projection, scored on the same players but left out of the ranking. Each source was evaluated on 98 QB, 199 RB, 197 WR and 98 TE player-weeks.
Week by week, the FFA Average tracks closely with the top comparison source in each panel: the other source with the lowest MAE at that position over Weeks 1-5.
2026 Next to 2024 and 2025
The 2024 and 2025 figures use the same six sources and the same player pool (FFA Average’s top 20 QB, 40 RB, 40 WR and 20 TE each week, where all six had a projection). The gap is largest at TE (+1.32 over 2024, +1.44 over 2025) and smallest at QB against 2024 (+0.33).
Part of that TE increase reflects scoring: TEs in the pool averaged 11.5 DK points per player-week, up from 8.3 in 2024 and 10.8 in 2025 over the same weeks. Scaled to projection size, the 2026 TE projections still missed by about 6.0 points for every 10 projected, compared with 5.5 in 2024 and 5.1 in 2025.
| Miss per 10 projected pts, Weeks 1-5 | QB | RB | WR | TE |
|---|---|---|---|---|
| 2024 | 3.38 | 4.61 | 5.33 | 5.55 |
| 2025 | 3.13 | 4.56 | 4.82 | 5.12 |
| 2026 | 3.45 | 4.84 | 5.72 | 6.00 |
2024 Weeks 1-16 and 2025 Weeks 1-15 give a broader baseline, and 2026’s five weeks are above them too:
| Six-source mean MAE | QB | RB | WR | TE |
|---|---|---|---|---|
| 2024, Weeks 1-16 | 6.05 | 6.14 | 7.08 | 5.36 |
| 2025, Weeks 1-15 | 6.04 | 6.20 | 6.82 | 5.41 |
| 2026, Weeks 1-5 | 6.45 | 6.40 | 7.79 | 6.46 |
Are Early Weeks Different?
In 2024 and 2025, Weeks 1-5 had a lower MAE than the later weeks in six of eight season-position comparisons. The exceptions were 2024 QB (6.11 vs 6.02) and 2024 WR (7.36 vs 6.95). Two seasons don’t establish a pattern, but these differences support comparing 2026’s first five weeks with the same weeks in prior years, while using the longer samples for context.
Why 2026 has been harder to project isn’t exactly clear from these results. Early-season uncertainty around depth charts, rookie usage and offseason changes occurs every year, so it alone wouldn’t explain a larger miss this season. Projections ran low at every position in 2026, as they did in Weeks 1-5 of 2025. This year’s misses were larger, and actual scores were more spread out than in the same weeks of 2025.
Players Behind the Numbers
Biggest Misses
Among players with at least three weeks in the pool, FFA Average’s largest net miss was Kenneth Walker. It underprojected him by 54.3 points over four weeks, or 13.6 per week. Saquon Barkley was the largest the other way: it overprojected him by 42.1 points (10.5 per week). Net totals let a miss in one direction offset a miss in the other, so they can understate how far off individual weeks were.
In single games, FFA Average’s largest underprojection was Tetairoa McMillan in Week 4 (51.2 points vs 15.5 proj), and its largest overprojection was Ja’Marr Chase in Week 1 (3.2 on 21.3). The five largest underprojected weeks all missed by more than 25 points, while no overprojection exceeded 18.1.
Sources sometimes split widely on the same player. In Week 4, ESPN projected Emanuel Wilson highest of the six at 14.2 and FantasySharks lowest at 5.2. He scored 27.0, and ESPN was closest.
Players Sources Kept Picking
A source “picks” a player when its projections put him in that source’s optimal DraftKings lineup. The pick statistics throughout this article use lineups with likely inactive players replaced, following the same procedure described in the lineup section below. These are projection optimized lineups to fit salary constraints, not public ownership. A player picked by five sources in one week counts as five picks but one game. A pick reaches 3x salary when the player scores at least 3 DK points per $1,000 of salary. A $6,000 player needs 18.
Amon-Ra St. Brown was the most frequent pick: 14 picks across all five weeks, and each of the six sources selected him at least once. The most-picked players reached 3x salary in half or fewer of their picked weeks, Kenneth Walker (2 of 3) was the exception. With two to five weeks per player, these counts are too small to separate one player from another.
| Source | Most-picked players, 2026 (picks) | 3x rate |
|---|---|---|
| CBS | Gibbs (3), Bijan (2), Christian Watson (2), Goedert (2), London (2) | 38% |
| ESPN | Aaron Jones (3), St. Brown (3), Jeanty (3), Adams (2), Price (2) | 40% |
| FFA Average | Gibbs (3), St. Brown (2), Brissett (2), Nacua (2), LaPorta (2) | 40% |
| FFToday | St. Brown (5), Josh Allen (3), Walker (3) | 35% |
| FantasySharks | Gibbs (3), St. Brown (2), McCaffrey (2), London (2), Kittle (2) | 38% |
| numberFire | Henry (4), Achane (2), Garrett Wilson (2), Murray (2), JSN (2) | 40% |
Figure 6 shows whether each source’s picks scored above or below their projections. Blue dots show the average projection error for picks. Black diamonds show the same measure for comparable players in that source’s pool. Left of zero means players scored more than projected. Right of zero means they scored less.
Five of the six sources’ picks landed right of zero, scoring below projection by 0.20 (ESPN) to 2.43 points (FantasySharks) per pick, while their comparable pools came out closer to projection or above it. FFA Average was the exception: its picks scored 0.27 points above projection, slightly less than its pool (0.38 above); in 2024 and 2025 its picks did better than its pool.
Mean error shows direction. A negative value means a source projected below what players scored. Looking at a source’s whole player pool and at only its picks can give opposite answers.
In 2026, ESPN underprojected every position and had the most negative average error at each. Its largest gap was at TE, where players scored 2.46 points above projection per player-week. FantasySharks went the other direction at QB, overprojecting by 1.64 points, while FFToday overprojected WRs by 0.72.
ESPN shows how the two can split. Across the shared pool its projections ran low, yet its optimizer picks scored 0.20 points below projection on average. The optimizer favors players a source projects highest relative to salary, which tends to include the ones it overrated. Overprojections and underprojections can also cancel, so an average near zero does not necessarily mean individual projections were accurate.
Player Stretches
We can also follow individual players over several weeks to compare projected points, actual scores and optimizer selections. They show how stretches of strong or weak scoring lined up with changes in projections and selection frequency. This helps explain how a player could attract picks after a strong stretch without delivering the same value in subsequent weeks.
These examples use FFA Average projections. Pick counts reflect selections across all six sources.
Rico Dowdle (2025) He scored more than 30 points three times in five weeks (41.4, 39.9 and 31.1 in Weeks 5, 6 and 9), each on a projection of 16.5 or less. From Week 10 his projection sat at 16.5-17.3 and one or two sources picked him each week, but he stayed under 3x salary.
Amon-Ra St. Brown (2026) He was projected between 20.2-21.7 every week and picked by at least one source every week. He scored above that range in Weeks 1 and 2 (28.7 and 38.2) and below it in Weeks 3-5 (11.9, 15.5 and 12.9).
Ja’Marr Chase (2025) He averaged 33.7 points in Weeks 5-7 on an average projection of 16.2, while no source picked him. His projection then averaged 21.5 over his next three games, and he averaged 15.7. Sources picked him in Weeks 8, 9 and 11, reaching 3x salary in none of those weeks.
Rachaad White (2025) The reverse case: he averaged 9.0 points in Weeks 7, 8 and 10 on a 15.9 projection, then 6.4 over Weeks 11-13 as his projection fell to an average of 10.2.
Which Source’s Lineup Scored the Most?
Each lineup starts as the optimal DraftKings lineup FFA’s optimizer builds from one source’s projections. If a player in it didn’t play, he’s swapped for that source’s next-best projected player at the same slot who fits under the $50,000 cap, and the rest of the lineup stays as built. The same lineups are used for the picks earlier in the article.
| Week | Lineup | Didn’t play | Replaced by | Replacement points |
|---|---|---|---|---|
| 2 | ESPN | Puka Nacua | Ja’Marr Chase | 26.5 |
| 2 | FFA Average | Zay Flowers | Rashee Rice | 12.3 |
| 3 | App DK | Jayden Reed | Xavier Hutchinson | 8.5 |
| 3 | ESPN | Jayden Reed | Malik Washington | 11.1 |
| 3 | FFA Average | Jayden Reed | Xavier Hutchinson | 8.5 |
| 4 | App DK | Jalen Coker | Stefon Diggs | 8.5 |
| 4 | App DK | Mason Taylor | Gunnar Helm | 3.5 |
| 4 | ESPN | Mason Taylor | Tommy Tremble | 5.2 |
| 4 | FFA Average | Jalen Coker | Stefon Diggs | 8.5 |
After subs, ESPN’s lineups averaged the most actual points in 2026, 148.7 per slate, ahead of the FFA app’s lineup (143.2) and FFA Average (141.8). Across the available 2024 Weeks 1-16 and 2025 Weeks 1-15, among nine sources, the FFA app’s lineup ranked second and sixth. The FFA Average lineup ranked third and fourth.
The takeaway five weeks in is a larger miss at every position, rather than a gap between providers. The lineup results are a separate view and don’t measure projection accuracy.
