Over 3,970 team-games since 2019, the defenses ESPN and RotoWire projected in a week’s top 10 finished in the top 10 about half the time, 49%. Picking by a defense’s season average got 41%, and picking at random about 37%. The projections help, in every season, but defense is the noisiest slot in fantasy, and points allowed is the part nobody predicts well.
Most managers stream their defense, so the weekly projection is often the whole decision. I graded eight seasons of it: every ESPN and RotoWire defense projection from 2019 through Week 10 of 2026, scored the same way and set against what each defense actually scored.
Update, Nov 21, 2026: ESPN and RotoWire fantasy football projections graded since 2019: they miss by nearly the same, and when they disagree, the lower one wins 58% of the time. Follow-up: Which NFL fantasy projections are better, ESPN or RotoWire?.
How well the projections rank defenses
A ranking is what you use on Sunday, so that’s the test: within each week, how closely does the projected order match the order defenses actually finished in? A rank correlation of 1 is a perfect order, 0 is a coin flip.
| Rank correlation | |
|---|---|
| ESPN and RotoWire averaged | 0.28 |
| RotoWire | 0.27 |
| ESPN | 0.25 |
| Season average | 0.10 |
| Last season | 0.10 |
The projections sit around 0.25 to 0.28. That’s far from perfect, but it’s two to three times better than going by a defense’s average so far this season (0.10) or last season’s (0.10). The order they give you is worth using.
| Finished top 10 | |
|---|---|
| ESPN and RotoWire averaged | 50% |
| ESPN | 49% |
| RotoWire | 48% |
| Season average | 41% |
| Random | 37% |
Put another way: each week, about 4.9 of the projected top 10 finished in the top 10. Going by the season average, about 4.1 did, and random picks land about 3.7.
It holds every season
- Projections
- Season average
The projections beat the season average every season from 2019 to 2026. The edge was smallest in 2019 and largest in 2023, when the season average did no better than chance.
What that means on Sunday
- The top of the board is worth chasing. The five defenses projected highest each week scored 8.9 points a game on average. The five projected lowest scored 4.2.
- Fade the bottom harder than you trust the top. The lowest-projected defense finished in the week’s bottom 10 in 61% of weeks. The highest-projected one finished in the top 5 in only 35%.
- Expect noise. A defense scores 6.4 points a game on average, but with a spread of 5.8. A quarter of team-games reach 10 points and 8% go negative.
- Points allowed is the hard part. The projected points allowed missed by about 7.3 to 7.5 points a game, about one full scoring tier.
ESPN or RotoWire?
It barely matters. RotoWire’s order was a little better in seven of the eight seasons (0.27 to 0.25 overall; ESPN’s in 2022), and its miss a little smaller (4.31 points to 4.35). But ESPN’s top 10 finished in the top 10 just as often (48.9% to 48.4%). Averaging the two ranked best overall, 0.28, though not every season. If both are in front of you, use the average.
| Method | Average miss | Rank correlation | Top 10 finished top 10 |
|---|---|---|---|
| ESPN and RotoWire averaged | 4.31 | 0.28 | 49.5% |
| RotoWire | 4.31 | 0.27 | 48.4% |
| ESPN | 4.35 | 0.25 | 48.9% |
| ESPN’s own total | – | 0.27 | 48.3% |
| RotoWire’s own total | – | 0.27 | 47.7% |
| Season average | 4.85 | 0.10 | 41.0% |
| Last season | 4.65 | 0.10 | 41.1% |
How CUTLINE scores defenses
CUTLINE’s contests score a defense on its impact plays: sacks, interceptions, fumbles recovered and defensive touchdowns. Points allowed doesn’t count. The story is the same there: over the same seasons, the projections ranked defenses at 0.24, against 0.09 for the season average. That’s why Counterpick’s defense board and Callout’s defense panel rank by the ESPN and RotoWire projections averaged.
For matchups by position, the fantasy points allowed tracker has every defense, updated each week.
How this was computed
- Scoring: standard DST, the common league default: sack 1, interception 2, fumble recovered 2, safety 2, blocked kick 2, defensive or return touchdown 6, and points allowed 0: 10, 1–6: 7, 7–13: 4, 14–20: 1, 21–27: 0, 28–34: −1, 35+: −4. Points allowed is the opponent’s final score.
- Projections: each source’s line for each week, rescored in that scoring from its own projected sacks, takeaways, touchdowns, safeties, blocked kicks and points allowed (each site’s own scoring differs a little). Graded on the 3,970 team-games where both had a line, 2019 through Week 10 of 2026.
- RotoWire: its projections come from Sleeper’s projection feed, which labels every line RotoWire.
- Rank correlation: within each week, how closely the projected order matched the actual one, averaged over 133 weeks.
- Top 10: of each week’s 10 highest-projected defenses, the share that finished in the week’s top 10 (ties at 10th count).
- Season average: a defense’s average score that season before the week, else last season’s.
Data: game stats from nflverse; projections from ESPN, and from RotoWire as Sleeper serves them.



