Defense rankings — 2026 Week 8
Use this for the order, not the number. Ranking defenses this way is about twice as good as ranking them by season-to-date fantasy points (Spearman 0.29 out of sample vs 0.13) — but half of what a defense scores is turnovers and touchdowns, which nothing predicts.
| Defense | Proj | Floor–ceiling | Pts allowed | from tiers | Sacks | Turnovers | |
|---|---|---|---|---|---|---|---|
| 1 | PIT vs CLE | 8.4 | 2.1 – 15.6 | 17.1 | +1.9 | 2.9 | 1.4 |
| 2 | DAL vs ARI | 8.1 | 1.8 – 15.2 | 18.6 | +1.5 | 2.8 | 1.5 |
| 3 | GB vs CAR | 7.6 | 1.3 – 14.7 | 19.5 | +1.3 | 2.7 | 1.4 |
| 4 | NYJ vs LV | 7.1 | 0.8 – 14.3 | 19.9 | +1.1 | 2.6 | 1.3 |
| 5 | LV @ NYJ | 6.7 | 0.4 – 13.8 | 21.4 | +0.8 | 2.5 | 1.3 |
| 6 | CLE @ PIT | 5.8 | -0.5 – 13.0 | 23.2 | +0.3 | 2.4 | 1.2 |
| 7 | BUF vs BAL | 5.6 | -0.7 – 12.7 | 24.7 | -0.1 | 2.2 | 1.3 |
| 8 | BAL @ BUF | 4.8 | -1.5 – 11.9 | 27.2 | -0.7 | 2.1 | 1.3 |
| 9 | CAR @ GB | 4.7 | -1.6 – 11.8 | 26.7 | -0.5 | 2.1 | 1.2 |
| 10 | ARI @ DAL | 3.7 | -2.6 – 10.8 | 29.5 | -1.1 | 1.8 | 1.1 |
Projected points allowed comes from the posted total and the opponent's implied
total; the tier column is the expectation across the points-allowed tiers, not the
tier a single projection happens to land in. Defensive touchdowns and safeties are
the league mean — nothing forecasts them
(fitted 2019-2025).
Measured walk-forward, one season held out at a time: Spearman
0.295, MAE 4.27 against a
4.41 league-mean baseline, R² 0.079
over 3,136 defense-games. The R² is the honest ceiling: this
orders defenses, it does not predict their scores. Study at
backtests/kicker_dst/FINDINGS.md.