> ## Content Index
> Fetch the complete content index at: https://www.statsdigest.com/llms.txt
> Use this file to discover other available public pages before exploring further.

# How Jets at Lions in Week 3 Actually Gets Decided
- URL: https://www.statsdigest.com/jets-lions-week-3-2026-quantitative-preview/
- Published: 2026-09-25T16:33:07.000Z
- Updated: 2026-09-25T16:33:07.000Z
- Description: Detroit's defensive collapse vs Buffalo and the Jets' surge vs Green Bay both reflect schedule, not skill.
- Author: Bill Seifert
- Tags: Jets, Lions, EPA, NFL Week 3, opponent adjustment, betting analysis

**NYJ at DET, Sunday Week 3, 2026\. Market line: DET −6.5, total 47.5.**

Two games is not a season. Everything below is written against a 2‑game sample per team — 135 offensive plays for Detroit, 131 for the Jets — and the schedules those plays came against are wildly uneven. Read this as an argument about *what would have to be true* for the game to break each way, not as a verdict.

## Table of Contents

1. What each team has actually done through 2 games
2. Opponent adjustment: how much of that is schedule?
3. A depth-of-target matchup, sized against the 2025 sample
4. Run game and what to do with it

---

## 1\. What each team has done, with the play counts attached

Through 2 games, per `team_weekly_epa`:

| Team | Side    | Week 1 EPA/play (plays) | Week 2 EPA/play (plays) |
| ---- | ------- | ----------------------- | ----------------------- |
| DET  | Offense | +0.00 (73)              | +0.23 (62)              |
| DET  | Defense | −0.03 (86)              | +0.43 (66)              |
| NYJ  | Offense | +0.17 (60)              | −0.11 (71)              |
| NYJ  | Defense | −0.12 (49)              | −0.22 (50)              |

Two things jump out immediately, and both need context before anyone leans on them. Detroit's defense allowed +0.43 EPA/play across 66 plays in Week 2 — that is a very bad number. And the Jets' defense has posted −0.12 and −0.22 EPA/play allowed, which through two games are elite figures. The league average sits at essentially 0.00 EPA/play on both sides in 2026 so far (3,835 plays either way).

The 2025 baselines say to be careful with both readings. Over full 2025 seasons (`team_season_epa`, \~1,000+ plays per side):

- **Detroit 2025:** offense +0.08 EPA/play, defense −0.01 EPA/play. A good offense, a roughly average defense.
- **Jets 2025:** offense −0.13 EPA/play, defense +0.15 EPA/play. A bad offense, a bad defense.

So the Jets' defensive numbers through two games are a large departure from a 1,053-play 2025 baseline, and Detroit's Week 2 defensive collapse is a large departure from its own 1,039-play 2025 baseline. Which brings us to the opponents.

## 2\. Opponent adjustment — the most important section

You cannot read the two-game numbers without knowing who each team played. Here's what each opponent has done across all of *its* 2026 games:

| Opponent (of…) | Their offense EPA/play | Their defense EPA/play |
| -------------- | ---------------------- | ---------------------- |
| NO (DET wk 1)  | +0.05 (150 plays)      | +0.05 (128 plays)      |
| BUF (DET wk 2) | **+0.34 (118 plays)**  | +0.14 (135 plays)      |
| TEN (NYJ wk 1) | −0.02 (100 plays)      | +0.10 (132 plays)      |
| GB (NYJ wk 2)  | **−0.20 (117 plays)**  | −0.07 (130 plays)      |

League average is 0.00 on both sides. So:

- **Detroit's defensive Week 2 line was against Buffalo, which is currently the most efficient offense in the league by EPA/play through two games.** That +0.43 EPA/play allowed is not a random data point — it is a data point about facing Josh Allen and the Bills' offense. It is still bad, but it is the kind of bad every defense is going to look against that opponent right now.
- **The Jets' defensive Week 2 line was against Green Bay, which is currently the *least* efficient offense in the league by the same measure.** That −0.22 EPA/play allowed is real production, but the opponent was giving points away everywhere it played.

Symmetrically on offense: Detroit's +0.23 EPA/play in Week 2 came against a Buffalo defense running +0.14 EPA/play allowed (bad). The Jets' −0.11 came against a Green Bay defense at −0.07 (good). Both offensive readings flatter or punish the team more than the underlying play deserves.

The honest one-line version: **through two games, we largely have a schedule comparison, not a team comparison.** The Jets' defensive breakout is against soft passing offenses; Detroit's defensive number is Bills-inflated. Both distortions will shrink with each additional game and are simply not resolvable at n=2.

## 3\. Depth of target — where the 2025 sample is more useful

The 2025 pass-defense-depth splits are 8× the sample of anything from 2026 and are the right place to look for a specific, measurable matchup. Against the 2025 league averages of +0.42 EPA/play allowed on deep throws (20+ air yards), +0.38 on intermediate (10–19), +0.12 on short (0–9), and −0.20 behind the LOS:

**Detroit's 2025 pass defense by depth:**

- Behind LOS: −0.34 EPA/play on 75 plays (better than league)
- Short: +0.02 on 268 plays (better than league)
- Intermediate: **+0.38 on 127 plays (league-average)**
- **Deep: +0.58 on 85 plays (worse than league)** — 4 INTs, 36.5% completion allowed

**Jets' 2025 pass defense by depth:**

- Behind LOS: 0.00 EPA/play on 100 plays (better than league)
- **Short: +0.39 on 264 plays (much worse than league +0.12)**
- **Intermediate: +0.51 on 99 plays (worse than league)**
- Deep: +0.50 on 51 plays (roughly league)

So the vulnerabilities point different directions. Detroit's 2025 weakness was over the top; the Jets' 2025 weakness was underneath and in the middle.

Now the important part — does the other team actually attack that zone? Through two 2026 games:

| Offense         | Attempts | Behind LOS | Short (0–9) | Intermediate (10–19) | Deep (20+) |
| --------------- | -------- | ---------- | ----------- | -------------------- | ---------- |
| DET 2026        | 77       | 8 (10%)    | 44 (57%)    | 20 (26%)             | 5 (6%)     |
| NYJ 2026        | 65       | 13 (20%)   | 36 (55%)    | 11 (17%)             | 5 (8%)     |
| DET 2025 (full) | 582      | 118 (20%)  | 303 (52%)   | 122 (21%)            | 39 (7%)    |
| NYJ 2025 (full) | 494      | 79 (16%)   | 285 (58%)   | 85 (17%)             | 45 (9%)    |

The Jets' 2026 sample is 65 attempts and looks like their 2025 shape, with a slight tilt to behind-the-LOS work. **The Jets are not a team that lives on 20+ air-yard throws (8% in 2026, 9% in 2025).** So Detroit's clearest 2025 weakness — deep — is a zone the Jets rarely target. If Detroit's secondary gets hurt in this game, it is more likely to be on the 5–15-yard stuff, where the Lions were only average in 2025.

Detroit going the other way is more actionable. The Jets' 2025 weakness was short and intermediate throws, and Detroit ran 57% of its 2026 passes short and 26% intermediate — 83% of its dropbacks in the exact zones where the Jets' 2025 defense gave up +0.39 and +0.51 EPA/play. That is the cleanest specific matchup in the game on paper.

Caveats: the Jets' defense has clearly been getting stops in 2026 (−0.12 and −0.22 EPA/play allowed), so either the personnel is different, the scheme is different, or the opponent quality is doing all the work. On the sample we have, I'd lean on the opponent-quality explanation given the Titans/Packers offensive profiles above, but it is unknown how much of the 2025 depth pattern still describes this defense.

## 4\. Run game and situational

**Rushing, 2026:**

- DET offense: 53 rushes, 4.36 YPC, −0.09 EPA/rush, 9 runs of 10+
- NYJ offense: 66 rushes, 3.32 YPC, −0.21 EPA/rush, 5 runs of 10+
- DET run defense: 61 rushes faced, 4.82 YPC, **+0.13 EPA/rush allowed**
- NYJ run defense: 32 rushes faced, 4.09 YPC, **−0.20 EPA/rush allowed**

The Jets have run the ball a lot and gained very little from it. Detroit's run defense number is ugly through two games — and again, one of those games was Buffalo. The Jets' run-defense figure comes off only 32 carries faced, which is a small sample even by 2026 standards; opponents have thrown against them 65 times to just 32 rushes, likely because both games were competitive/trailing scripts for the offense.

If this game plays script — Detroit favored at home — the Jets' run-defense number becomes less relevant because the Lions will not need to lean on rushing volume. If the Jets keep it close early, we get a better test of it.

## What would have to be true for it to break each way

**For Detroit to cover −6.5:** the Week 2 offensive line against Buffalo's defense (+0.23 EPA/play) travels against a Jets defense whose two-game numbers are largely a soft-schedule artifact, and Detroit lives in the 5–15-yard passing zones where the Jets were worst in 2025\. That is a coherent story from the data.

**For the Jets to cover:** the defensive improvement is real and not a schedule illusion — meaning the Titans/Packers offensive quality above understates the Jets — and the Detroit offense regresses back toward its 0.00 EPA/play Week 1 line rather than its +0.23 Week 2 line. That is also coherent, but requires believing that the largest signal in the two-game data (opponent quality) is not the main driver.

At a two-game sample against opponents this uneven, the market's −6.5 is the most informed prior available and neither the descriptive projection nor the blessed model gives a good reason to overrule it. The interesting football question — whether the Jets' defense is genuinely a different unit than in 2025 — will not be answered by this game alone.