Four Upsets, None of Them Ours

NFL 2026  ·  Week 7 in review  ·  Thursday Oct 22 – Monday Oct 26

Contents

01  The Week in One Line

02  The Scorecard

03  Game by Game

04  What We Got Wrong

05  What We Got Right

06  The Week That Was

07  Injuries That Change Things

08  The Season So Far

A winning week that still lost to the market: 9 right, 5 wrong, one game worse than simply taking the betting favorite in every game (10-4). Four underdogs won outright — Atlanta, Tampa Bay, the Giants and Miami — and we had the favorite in all four, two of them at 70% or more. Our one pick against the market, Kansas City at Seattle, lost 20-3.

The winners column flatters the week. Our average miss on the final margin was 10.8 points, the forecasts were wrong about the shape of most games, and the two calls we got exactly right on margin, Steelers by 6 and Lions by 3, both missed the total by 14 points. Over the season we are now seven games behind the favorites, 61-46 against 68-39.

↑ Back to contents

The margin misses were large and lopsided. Three games went off by 19 points or more: Tampa Bay’s 43-13 win in a game we had Carolina winning by four (a 34-point miss), Seattle’s 20-3 win over our pick Kansas City (20), and Cincinnati’s 35-13 win in a game we had as a three-point Bengals win (19). Against the spread we went 5-7; two games had no lean because we printed a 50% cover chance.

On scoring, our total-points forecasts ran 1.2 points low on average, but that average hides a split week. Four games finished at least 14 points under our number — Paris (26 against 40), Colts-Vikings (26 against 40), Packers-Lions (37 against 51) and Chiefs-Seahawks (23 against 43) — and four finished at least 14 over. Eight of fourteen went the way of the over/under lean we printed.

↑ Back to contents

Every game, in the order it was played: how it was won, who won it, and what we had called before kickoff.

At a glance

↑ Back to contents

Buccaneers 43, Panthers 13 — a 34-point miss. We had Carolina by four, at home, against a Tampa team we described as a bounce or two from 3-3. We expected Tampa’s run defense to be the story and it was its run offense: 258 yards on 48 carries against a Carolina front missing Jaelan Phillips. We built the forecast around Mayfield’s return; he threw 11 passes. Part of this is variance — a defensive touchdown and a two-point return on a blocked extra point are not forecastable — but Tampa led comfortably before either, and a game that ends 43-13 was not a coin flip that landed the other way. We misjudged the matchup.

Seahawks 20, Chiefs 3 — our one pick against the market, wrong by 20. We took Kansas City at 59% because we trusted Kenneth Walker running at Seattle’s depleted linebackers more than we trusted Seattle’s offense. Walker ran for 36 yards and the Chiefs scored three points. The Thing To Watch named the real danger — Kansas City’s backup tackles in the loudest stadium they will play in — and then the pick ignored it. That makes four wins in fifteen picks against the favorite this season, when our own stated probabilities said to expect about nine.

49ers 20, Falcons 29 — 70% on the loser. We said the market had moved too far toward Atlanta on the 49ers’ injuries. San Francisco outgained Atlanta by 105 yards and lost because it gave the ball away four times. Turnovers in a game that was otherwise going to plan are mostly variance; this one we would make again at a lower number.

Giants 27, Texans 21 (OT) — 77% on the loser. Our most confident call of the week. We wrote off a first-time starter behind backup tackles; McCarthy completed five throws of 20-plus yards and Skattebo ran for 124. The forecast landed half a point short of the 10.5-point market, so this was the market’s miss as much as ours, but we priced it at three wins in four and it lost.

Dolphins 28, Jets 25 — 70% on the loser, 14 points low on the total. We trusted the Jets’ defense against Miami’s passing game and predicted 39 total points; the game produced 53, and Miami got 103 rushing yards from a backup back. The Jets’ red-zone failure we flagged in advance was what beat them.

Bengals 35, Ravens 13 — right winner, 19-point miss. We had a three-point game. It was a 13-0 Baltimore lead and then a 35-0 Cincinnati half. Picking the winner is not the same as reading the game, and we did not read this one.

↑ Back to contents

Nine winners, and every one of them was the betting favorite. Where we disagreed with the market, we lost.

The closest calls were Steelers by six in Paris (exactly six, though the total missed by 14), Lions by three (exactly three, and the total missed by 14 the other way), Browns by four (by three) and Vikings by eight (by six). In each the spread was within a point or two of our number, so the credit is shared with the market.

Two pre-game reads held up on their own. In Bengals-Ravens we said that if Cincinnati held Derrick Henry under four yards a carry, Tyler Huntley would have to throw on second-and-long and get sacked; Henry averaged 2.6 and Huntley was sacked four times. In Colts-Vikings we said Minnesota’s blitz against Daniel Jones on third down was the matchup; the Vikings blitzed at a record rate and Jones threw the deciding interception.

Two of the four underdogs that won did it the way our Underdog’s Case described. For Miami we wrote that the Jets’ red-zone failures could hand a grinding Dolphins team a close game; the Jets kicked a field goal from the 1 and lost by three. For the Giants we wrote that Nabers against Houston’s depleted secondary, and a defense that kept the game close into the fourth quarter, was the path; that is roughly how it went, though Skattebo’s running, which we did not name, finished it. We still picked against both.

↑ Back to contents

The unbeatens are down to one. San Francisco and Kansas City both lost for the first time, leaving Minnesota at 6-0 — eleven straight wins counting last season, built almost entirely on a defense that held Indianapolis scoreless after halftime and an offense that threw for 67 yards. The Vikings are winning games like this every week, and at some point the scoring has to come from somewhere.

Two second-half explosions. Cincinnati was shut out for a half in Baltimore and then scored 35 points in 30 minutes, Burrow completing 16 of 18. Tampa Bay ran for 258 yards in Charlotte in a game that ended with Carolina’s starters on the bench. Both results change how those divisions look: Cincinnati is 4-2 and Baltimore’s season now depends on when Lamar Jackson’s ankle allows him back.

Running backs everywhere. Seven backs ran for 100 yards or more: Javonte Williams (155), Tony Pollard (129), Cam Skattebo (124), Bucky Irving (106), Ollie Gordon II (103), Jaylen Warren (101) and Emanuel Wilson (100). Dallas had not had a 150-yard rusher since 2018 and got one in Philadelphia.

Kickers decided three games. Andre Szmyt and Jake Bates won games with kicks at the gun; Cairo Santos hit a career-long 58-yarder to win Thursday’s game. On the other side, Green Bay’s Trey Smack missed one and had another blocked in a three-point loss, and Indianapolis’ Spencer Shrader missed twice in a six-point loss.

Philadelphia is in freefall. Five straight losses, Jalen Hurts out for the season, and a 38-21 home defeat to the team it least wants to lose to. Dak Prescott became Dallas’ all-time touchdown leader on the night. Both teams are below .500 in an NFC East where the Giants, at 4-2-1 after a first start from J.J. McCarthy, now lead.

↑ Back to contents

Jalen Hurts, Eagles (blood clots). Out for the season. Tanner McKee is the starter for a 2-5 team, and Saquon Barkley, back from a hamstring injury, left Monday’s game with a back injury; Dameon Pierce also left. Philadelphia’s running game is down to whoever is healthy on Wednesday.

Lamar Jackson, Ravens (ankle). Three missed games. Jesse Minter expects him to be “in a decent spot” for Week 8, with more on Wednesday. With Huntley, Baltimore has scored 10 and 13 in its last two games; his status is the biggest number on next week’s Ravens line.

Taylor Moton, Panthers (fibula/ankle). On injured reserve after Sunday’s loss; NFL.com reports season-ending ankle surgery, while Dave Canales said only that he will be out “a while.” Carolina, already missing Jaelan Phillips and two other defensive starters at practice, loses its veteran right tackle.

Charles Cross, Seahawks (high ankle sprain). Seattle hopes for one to two weeks. Rashid Shaheed hurt his knee in a collision with Smith-Njigba; early returns were described as positive.

Christian McCaffrey, 49ers (hamstring). A mild strain, and San Francisco has a bye. Kyle Shanahan said the team is optimistic he will be fine after it.

Green Bay’s offensive line. Center Sean Rhyan is believed to have a high ankle sprain, a type of injury that often costs four to six weeks; Jacob Monk went on injured reserve; Aaron Banks and Travis Glover are also hurt. Green Bay signed Mekhi Becton from the practice squad. Receiver Matthew Golden (shoulder, foot) did not practice Monday.

Jaylen Waddle, Broncos (chest). Hurt diving on a fumbled reverse late in Arizona; severity not yet known. Sean Payton took the blame for the call.

Also: Aaron Jones, Vikings (knee, day-to-day); Cam Skattebo, Giants (hamstring, expected to be fine, and New York is on bye); Arizona receiver Jalen Brooks (season-ending wrist surgery); Baltimore lineman Olaivavega Ioane (knee); Justin Reid, Saints (concussion); Jaylin Noel, Texans (turf toe).

↑ Back to contents

Seven weeks in, we are 61-46 picking winners (57%). Taking the favorite every time would be 68-39 (64%). Week 4 remains the only week we have finished ahead of the favorites; the gap is now seven games.

Picking against the market is the clearest problem, and after this week it is no longer a small-sample excuse. Those picks are 4-15 over the season, made at 53–62% confidence, when about nine wins would have been expected. Starting Thursday, any forecast that picks the underdog to win outright will have its predicted margin pulled 1.5 points toward the betting line before the probabilities are read off it. Some of those picks will flip back to the favorite; the ones that survive will be stated with less confidence.

Our least confident picks are still winning less than half the time, 19 of 41. The 60–69% band is on its stated rate. The 70–79% band went 2-3 this week and is now at 63% over 35 games, about ten points below what we said; that is within normal variation for 35 games but it is the second straight week it has slipped. The 80% band has only two games and says nothing yet.

Games started by a quarterback who did not start Week 1 — Penix, McCarthy, Huntley and McKee this week — went 2-2, with an average margin miss of 16 points against 10.8 for the week. Over 13 such games this season we are 3-10 and miss by about 13 points on average, against 9.9 overall. Two more games and the sample is large enough to act on.

On scoring, our forecasts ran 1.2 points low this week. Over 107 games they are 2.3 points low on average; over the 43 games since we raised every total by 1.5 points, 1.3 low, which is within normal variation. No further change to totals yet.

Not betting advice. The forecasts graded here are a model’s estimates and will be wrong in both directions; they are published so they can be scored, and this page is the scoring.

↑ Back to contents