Seattle won by 24. The line everyone suspected was wrong wasn't.

Seattle beat Arizona 31-7 on September 20. That closes the book on a spread dispute this site wrote about three days before kickoff — and the way it closed inverts the piece's own framing of which number to distrust.

What the September 17 piece actually said

SEA@ARI had carried a real spread disagreement for four straight days: every source had Seattle favored since September 15, but by how much ranged from GameScript's own house line at -3.5 up to a reported DraftKings number of -10 — a 6.5-point gap on a single game, three days out. The piece named the house number as the one worth flagging specifically: it was the newest entrant, it didn't side with either cluster, and it landed as the low outlier on the opposite end from DraftKings. The total, by contrast, had stayed tight all week, 40.5-41.5 across every source.

The rule the piece closed on: don't average a 6.5-point gap and call it a lean, and treat it as unresolved until a source or two closes it.

What actually closed it

DraftKings' own board is what moved. By kickoff, DraftKings had Seattle -3.5 at -120, total 40.5 (FantasyAlarm's odds page), with FanDuel sitting a point off at -4.5/+4.5 and a 41.5 total (SI.com, FanDuel's own research page, DeFiRate). That's a tight, converged market — and it converged on GameScript's end of the range, not DraftKings' own three-days-out number. The -10 this repository's research had reconfirmed as late as September 17 was never where the game closed; whatever produced it had five more days to move, and it moved the rest of the way to meet the house line instead of the other way around.

That's worth sitting with, because the piece's own instinct — that a feed reading "closing consensus" days before anything has closed is the one to doubt — pointed at the wrong number here. The newest, most isolated-looking read was the one that held.

What neither number saw coming

None of that means the house line called the game. Seattle won by 24. Even DraftKings' widest, most-doubted three-days-out number, -10, implied a margin less than half that. Drew Lock, starting for an injured Sam Darnold with no update on that question ever surfacing that week, threw for 235 yards and 3 touchdowns. Jaxon Smith-Njigba caught 9 passes for 155 yards and all three of them, a stat line Rotoballer flagged afterward as proof he's the rare receiver who doesn't need a particular quarterback to produce. Arizona's offense, with Jacoby Brissett at quarterback, managed 95 passing yards and seven total points.

A -3.5-to-4.5 spread describes a competitive game. A 31-7 final is not that, by a wide margin. The September 17 piece framed the open question as "which shape is this game" — a low-total grind or a low-total blowout — and the honest answer is that it was neither of the shapes either candidate spread implied. It was a blowout inside a merely-ordinary total: 38 combined points landed close to the stable 40.5-41.5 range everyone had agreed on for a week, while the margin that produced those points was not something any number on the board, including the one that ultimately proved closest, priced in.

The actual lesson

The piece's rule — don't resolve a real spread disagreement by averaging it, wait for sources to converge — held up. What it got wrong was which side of the gap to bet would hold: the newer, more isolated number can simply be ahead of a line that hasn't finished moving yet, not a glitch to discount because it disagrees with a louder consensus. And the total's stability, which the piece treated as the more trustworthy half of the line, turned out to be right about that.

Worth naming directly: the total's job is to predict combined scoring, not game competitiveness, and it did its job here — 40.5-41.5 projected against 38 actual is a closer miss than any spread on the board managed against a 24-point margin. A settled total told you how many points were coming. No spread, settled or not, told you how lopsided the game splitting them would be.