> ## 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. # What Four Weeks of 2026 Actually Tells Us - URL: https://www.statsdigest.com/what-four-weeks-of-2026-actually-tells-us-week-4/ - Published: 2026-10-08T11:55:34.000Z - Updated: 2026-10-08T11:55:33.000Z - Description: Minnesota's defense and Dallas's are real; the 49ers' league-leading offense is schedule-inflated. - Author: Bill Seifert - Tags: NFL, EPA, sample size, 2026 NFL season, Vikings, Cowboys Through Monday, September 14 and the following three weekends, 64 of 272 regular-season games are in the books — 24% of the schedule. Teams have logged between 192 and 292 offensive and defensive plays apiece. Chicago, running the league's fastest no-huddle, leads with 292; Miami, playing the fewest snaps, has 209\. A full season for a team is typically closer to 1,000 plays per side. So any league-wide ranking you see this week is being read off a sample roughly a quarter the size of a full year's — and often less than that, because the top of a leaderboard is usually set by whoever drew the softest four opponents. That framing is the point of this piece. Below is a tour of what the first four weeks do, and do not, support. A companion article covers week 4 on its own. ## Table of Contents 1. The sample problem, quantified 2. Six things the first four weeks seem to be telling us 3. Three storylines where the move is biggest 4. Believe it or not yet --- ## 1\. The sample problem, quantified Teams have played 3 or 4 games. Per-side play counts for 2026 so far range from **209 (Miami) to 292 (Chicago)** on offense/defense combined weekly totals, versus \~1,000+ in a full 2025 season — Seattle alone ran 1,244 defensive plays in 2025\. That means a team's 2026 EPA/play is being computed off roughly 20–25% of a normal season's snaps, and against four specific opponents rather than seventeen. Opponent adjustment matters more than usual here. San Francisco's +0.29 offensive EPA/play currently leads the league, but two of its four opponents — Arizona and Miami — ranked **26th and 27th in 2025 defensive EPA/play** (+0.088 and +0.088). The other two (Denver 7th, the Rams 11th) are tougher — and against those two the 49ers posted +0.18 (week 1 vs LA) and +0.12 (week 4 vs DEN). Against Arizona and Miami they posted **+0.49 and +0.46**. The season-to-date aggregate is both numbers blended; treat it accordingly. Every rate in what follows is paired with its denominator for that reason. --- ## 2\. Six things the first four weeks seem to be telling us **1\. Minnesota's defense has been elite in all four games, not one.** The Vikings' cumulative defensive EPA/play is **−0.179 on 236 plays**, best in the NFL. The weekly numbers: **−0.175 (67 plays vs GB), −0.125 (69 vs CHI), −0.246 (64 vs TB), −0.170 (36 vs MIA)**. Two of those opponents (Green Bay and Chicago) ranked **3rd and 8th in 2025 offensive EPA/play**. In 2025 Minnesota was already the league's best defense at **−0.101 EPA/play on 1,016 plays**; this is that signal reinforced, not invented. **2\. Dallas's defense is historically bad again, and the opponents haven't been elite.** Dallas has allowed **+0.201 EPA/play on 240 defensive plays** — worst in the league. Weekly: **+0.369 vs NYG, +0.069 vs WAS, +0.134 vs BAL, +0.238 vs HOU**. Only one of those four opponents (Baltimore) was a top-tier 2025 offense. The 2025 Cowboys allowed **+0.167 EPA/play on 1,027 plays** (worst in the league). Four games in, 2026 looks like the continuation of a very large prior-season sample, not a fluke. **3\. The Chargers' offense has collapsed from a mediocre 2025 baseline.** Los Angeles is at **−0.159 offensive EPA/play on 245 plays**, dead last. In 2025 the Chargers were **−0.035 on 1,139 plays**. The weekly track: **−0.244, −0.261, −0.071, −0.085** — the first two games were the worst and the last two are closer to their 2025 level, which is itself a problem given that 2025 was already below league average. **4\. Philadelphia's offense is pointing in the wrong direction, and it's not a one-week thing.** The defending division's biggest 2026 surprise so far: **−0.107 offensive EPA/play on 228 plays**, 27th. The 2025 Eagles were **+0.021 on 1,067 plays**. More importantly, the weekly track is monotone bad: **+0.071, +0.039, −0.304, −0.287** — fine through two weeks, then two weeks of blowout-level negative splits. Two games is still two games, but it is not a single outlier. **5\. San Francisco leads the league in offense and third-down conversion, but caveat the schedule.** The 49ers have converted **23 of 40 third downs (57.5%)** through four weeks, best in the league. Buffalo (**24/44, 54.5%**) and Dallas (**26/49, 53.1%**) are the only other teams above 53%. Pair that with San Francisco's **+0.29 EPA/play on 222 offensive plays** and the opponent note from §1: two of the four defenses faced were bottom-six in 2025. **6\. Chicago's offensive breakout is one game plus three ordinary ones.** Chicago's cumulative offense is **+0.107 EPA/play on 292 plays** — top-nine. But the week-by-week EPA/play is **+0.409 (70 plays vs CAR), −0.125 (69 vs MIN), +0.093 (67 vs PHI), +0.058 (86 vs NYJ)**. Strip out the week-1 60-burger in Carolina and the Bears look like a competent, not elite, offense — up from 2025's **+0.071 on 1,253 plays** by a slim margin. --- ## 3\. Three storylines where the move is biggest ### San Francisco's offense: +0.06 in 2025 → +0.29 now, on 222 plays The 49ers were **+0.061 on 1,180 offensive plays in 2025**. Through four games of 2026 they are **+0.291 on 222 plays** — nearly five times the rate, on roughly a fifth the sample. The split is as revealing as the headline: | Week | Opponent | 2025 opp DEF rank | 2026 SF off EPA/play | Plays | | ---- | -------- | ----------------- | -------------------- | ----- | | 1 | at LA | 11 | +0.179 | 64 | | 2 | MIA | 27 | +0.460 | 47 | | 3 | at ARI | 26 | +0.490 | 49 | | 4 | DEN | 7 | +0.120 | 62 | I'd read that as "clearly better than 2025, probably not the best offense in football." The two monster games came against defenses that were below average a year ago. ### Minnesota's defense is doing exactly what it did in 2025, maybe more The Vikings allowed **−0.101 EPA/play on 1,016 plays in 2025** — the league's best figure. Through four weeks they are **−0.179 on 236 plays**, and they've done it against a mix that includes two 2025 top-ten offenses. If anything is sturdier than Minnesota's defense through four weeks, it is hard to find in this dataset. ### Dallas's defense: the problem has not moved Dallas was **+0.167 EPA/play allowed on 1,027 plays in 2025** (last). It is **+0.201 on 240 plays** in 2026 (last). Across those four games the Cowboys surrendered a weekly low of +0.069 and a weekly high of +0.369\. There is no four-week window in there that resembles league average. This isn't a 2026 story so much as a 2025 story that has refused to end. --- ## 4\. Believe it or not yet The question for each: is the four-week signal strong enough to update what I thought entering the season? - **Minnesota's defense is elite — believe it.** Already true in 2025 (1,016 plays), and the 2026 rate is better against a credible slate. What would kill the read: a stretch of giving up +0.00 or worse against mid-tier offenses. - **Dallas's defense is bottom-tier — believe it.** Two consecutive seasons of worst-in-league rates, and the 2025 sample alone is enormous. Confirmation requires nothing; falsification would need weeks of top-ten defensive EPA against real offenses, which has not happened yet. - **San Francisco is the best offense in football — not yet.** +0.29 is real and better than 2025, but half the sample came against two bottom-six-in-2025 defenses. I'd want to see another +0.20 game or two against a top-ten unit before calling it settled. - **Philadelphia's offense has a real problem — leaning yes, but it is two games.** The weekly track going from +0.07/+0.04 to −0.30/−0.29 is not noise-shaped; it is a step change. But a step change over two games is still two games. The next two will tell us a lot more than the previous four. - **The Chargers' offense is in trouble — leaning yes.** −0.159 cumulative is far below their 2025 baseline (−0.035, 1,139 plays), and even the "recovery" weeks (−0.07, −0.09) are below average. A couple of games near zero would meaningfully soften the read. - **Chicago's offense is a top-ten unit — not yet.** The aggregate +0.107 leans heavily on one 70-play explosion against Carolina (+0.41). The other three games averaged out closer to their 2025 baseline. A couple more weeks above +0.15 against competent defenses would make the case. --- A reminder of the frame: a 2026 number against a 2025 or multi-year baseline is a story; a 2026 number by itself is a sample. Where the sample confirms a large prior-year sample (Minnesota, Dallas), it is doing most of the work. Where it contradicts the prior (Philadelphia, LAC), it is small enough that I'd want another two or three weeks before treating it as settled. That is what four games — and only four games — can honestly tell us.