Skip to main content
The ledger · week 3, 2026

Week 3 graded: 7‑9, market 8‑8

The market beat us this week. Every pick was published before kickoff; here are the results, including the ones we got wrong.

29 Sept 2026 · 2 min read
7-9model, week 3

Every week we grade every pick we published, against the result and against the market. Week 3: the model went 7-9; the market favourite went 8-8. The market beat us this week.

7-9Model, week 344%
8-8Market, week 350%
0.270Brier scorelower is better; 0.25 is a coin flip
25-23Season recordmarket 65% of 48

Where we were wrong

SEA at 74% (SEA 31 @ WAS 33). LAR at 70% (LAR 26 @ DEN 30). GB at 70% (ATL 35 @ GB 14). The most confident misses are the ones worth learning from, so they go first.

Us vs the market

We were right where the market was wrong 0 times; the market was right where we were wrong 1 time (NYG).

One note on the market's number. It's graded at the odds when we published our picks, so the comparison is like for like: 8-8. At closing prices the market went 9-7. The difference is Rams at Denver: the Rams were 54% favourites when we published, and by kickoff Denver had closed as 1.5-point favourites. Denver won 30-26.

GamePickModel %ScoreModelMarket
LAC @ BUFBUF82%16-24RightRight
SEA @ WASSEA74%31-33WrongWrong
ARI @ SFSF70%30-36RightRight
LAR @ DENLAR70%26-30WrongWrong
ATL @ GBGB70%35-14WrongWrong
KC @ MIAKC69%24-10RightRight
NE @ JAXJAX66%6-35RightRight
LV @ NONO60%35-27WrongWrong
CAR @ CLECAR60%18-21WrongWrong
NYJ @ DETDET59%24-31RightRight
PHI @ CHIPHI57%7-27WrongWrong
BAL @ DALBAL54%34-31RightRight
TEN @ NYGTEN53%7-12WrongRight
MIN @ TBMIN52%23-16RightRight
CIN @ PITCIN52%27-30WrongWrong
HOU @ INDHOU52%17-19WrongWrong

Every week 3 pick, most confident first

Read this before you bet

Picking winners is not beating the price. Across 2025 our flat-bet ROI at posted odds was negative. Use the model as one view of the game, not as a tip.

Sources: results_dashboard.json (2026-09-29T09:00:48.715979Z). Every number here comes straight from our pipeline.