Skip to content

NFL Market

Week 1 · book vs exchange

GameBookExchangeModelGap
NESEA
-3.5
SFLAR
-0.5
NODET
-7o49.5-775%-7.5-0.03634k
NYJTEN
-1.5o38.5-0.752%-2.5+0.82309k
BALIND
+3.5o47.5+3.140%-3.5-0.41373k
CHICAR
+3o47.5+2.940%+1.5-0.09544k
CLEJAX
-8.5o39.5-8.979%-11.5-0.4428k
TBCIN
-3.5o50.5-3.664%-0.5-0.12296k
BUFHOU
+1.5o44.5+1.248%-2-0.31329k
ATLPIT
-6o40.5-6.172%-4.5-0.12830k
GBMIN
-2.5o46.5-2.155%-1.5+0.4760k
WSHPHI
-6o44.5-5.868%-8+0.23369k
ARILAC
-9.5o47.5-8.981%-8.5+0.6428k
MIALV
-3o40.5-3.161%+2-0.11304k
DALNYG
+3o48.5+3.140%-1.5+0.051087k
DENKC
-2.5o43.5-2.456%PK+0.14536k

14 of 16 games are quoted on the exchange. Spreads are home-relative — a negative number means the home team is favoured. Gap is the exchange minus the book on the same game: a disagreement between two venues, not a call by us. Model is ours — point differential, shrunk and adjusted for home field, fitted on 1,087 games and scored on seasons it never saw. There is no “edge” column because we measured for one and did not find it: against the closing line over a full season our number went 133–133, and the closing line predicted the final margin more accurately than we did.