Sign in
NFL
Write-ups
Write-upsPlayersBacktestHow this works

Week 1 review

Sep 15 · predictions are published to show the model works. nothing here is betting advice.Last updated: predictions Oct 1, 11:02 AM ET · results graded Sep 28, 11:22 PM ET
Written by our system from the week's data, with the same language model that reads the news; no human edited it. Every number comes from the Record and Week 1 pages. If a sentence reads wrong, tell us.

Week 1 in review

The model went 9–5 straight-up (64%) in Week 1. The market pick at kickoff performed better, going 11–3 over the same games. Against the market margin and on totals, the model finished 6–8 in each category.

The misses

Five games did not go as predicted:

  • ARI @ LAC: picked Los Angeles Chargers at 75%, final 26–14. The market agreed.
  • DEN @ KC: picked Denver Broncos at 62%, final 10–31. The market picked Kansas City instead.
  • MIA @ LV: picked Miami Dolphins at 57%, final 13–27. The market picked Las Vegas instead.
  • NYJ @ TEN: picked Tennessee Titans at 53%, final 23–10. The market agreed.
  • BUF @ HOU: picked Houston Texans at 51%, final 36–31. The market picked Buffalo instead.

The biggest surprise was the Chargers at 75%: they lost at home to Arizona by 12.

Where we differed from the market

The model disagreed with the market pick at kickoff in 4 games and went 1–3:

  • BUF @ HOU: model had HOU (51%), market had BUF (−1.5). Market was right.
  • MIA @ LV: model had MIA (57%), market had LV (−3). Market was right.
  • DAL @ NYG: model had NYG (55%), market had DAL (−3). Model was right.
  • DEN @ KC: model had DEN (62%), market had KC (−2.5). Market was right.

What the news layer did

Six injury and roster claims were applied before kickoff. Two pointed the right way; none flipped a pick.

The claims that helped: a Packers guard inactive for the Minnesota game, and a receiver elevated from the practice squad for Miami. The claims that hurt: three separate notes about Denver's undrafted rookie left tackle starting for the injured Josh Simmons, and a Dolphins tight end elevated to replace an injured player. One claim about a Las Vegas receiver also hurt.

Of the claims the model declined to use, eleven would have helped. The pattern is worth watching but not yet worth changing the rules.

Rankings after the week

  • Kansas City Chiefs up 4 to #14 (from #18).
  • Denver Broncos down 3 to #15 (from #12).
  • Cincinnati Bengals up 1 to #10 (from #11).
  • Green Bay Packers down 1 to #16 (from #15).

One week moves the ratings a little by design. The current season carries most of the weight, but that balance shifts toward recent games as the year goes on.

Next: Week 2

Picks are up on (/sports/nfl/) and update four times a day until kickoff.

Predictions are published to show the model works. Nothing here is betting advice.

← All write-ups