OddsRecord

Recorded before kickoff

The first forecast record

All 16 Week 1 forecasts in the public chain. None is scored yet because the games have not been played.

A probability is not a promise. A favourite can lose without the number being dishonest. Judge forecasts across a large sample, not by one game.
16forecasts recorded
7,276historical test games
10.1%Brier skill vs. base rate
1.1%calibration error
Kickoff (UTC)GameModel favoursWin probabilityRow hash
NE at SEASEA71.0%a28b4f51c5ef
SF at LALA66.6%9ba797934525
CHI at CARCHI61.6%b5a4f983e548
TB at CINCIN56.2%bfbdafcc13f5
NO at DETDET76.6%763c6925c333
BUF at HOUHOU56.2%70b2d6e23081
BAL at INDBAL57.4%70bdf585ce1d
CLE at JAXJAX80.2%fb2892b5bb37
ATL at PITPIT63.2%13b9eca95487
NYJ at TENTEN56.0%9dd3fa47bef7
ARI at LACLAC78.1%f263357e2679
MIA at LVMIA63.0%9838060ce800
GB at MINMIN61.0%7d91de1c65c4
WAS at PHIPHI73.3%064495fe8b50
DAL at NYGNYG51.1%b13f8ef88111
DEN at KCDEN63.9%ca94e6147c63

Chain tip: ca94e6147c63b307da7dc6bc0e8016adeebcf839c2fa43682351f07ee304ea59. Each row hash includes the prior row hash, so changing an old value breaks every later link.

What is measured and what is not

The 10.1% skill score and 1.1% calibration error come from a separate walk-forward test over 7,276 completed historical games. They describe that test. The forecasts in this table are the untouched live record and have no results yet.

No win rate, profit claim or selected streak appears here. When games settle, the original probability remains and the result is added beside it.