NFL Betting Desk · The Dynatyze Line
The Line's NFL Report Card
Every settled NFL moneyline gets graded against the close — the last pregame quote, the sharpest number a market ever prints. No cherry-picking, no deleted misses.
Does the number come true?
Our fair probability vs how often those bets actually hit.
Across 60 settled bets, our fair number lands inside its 95% band in 0 of 1 bucket carrying 25+ bets, and across all 5 buckets it sits 30.3pp off the diagonal, bet-weighted. Our heaviest bucket priced 35 bets near 54%; they hit 14% — 40pp from the number we quoted.
Every +EV bet we surfaced, graded win or loss against the game log — grouped by our fair probability, not cherry-picked. Settles where a box score can prove it: NFL.
50 NFL games graded so far — read everything below as direction, not verdict. A credible long-run read arrives around 100+ games.
The record, week by week
Click a column for that week’s detail · bar height = beat-close rate, bar width is fixed
W2 is the best week on the board: 75% of that slate's leans were confirmed by the close. A column is greyed when fewer than 8 games have settled — it is drawn, but never coloured as a verdict.
When The Line disagrees with the crowd
Bucketed by how much harder The Line leaned than the field at open.
- Avg CLV
- -1.62%
- Avg edge
- 0.4pp
- Avg CLV
- -10.20%
- Avg edge
- 1.4pp
Every settled game counted — confirmed or not.
Is The Line calibrated?
Predicted (open) probability vs where the market actually closed.
Not plotted: the 70–80% bin with 1 read. A bin needs eight reads before its closing figure is drawn — a verdict from one game is not a verdict.
Recently graded
What “the close” means here
On 41 games the close is now the book's own final number, stamped the moment it came down.
Games graded on the last number we saw before kickoff. Every grade above sits on this basis today.
Closing prices the book itself stamped, across 41 games. As these fill in, this record stops being graded against an estimate.