Lines · 2026 · Week 10 · read Fri 7:24 PM ET
299 lines posted.
The model leans on 156.
Every posted line this week, the market's price, the Second Opinion beside it, and what leans that size have done before. Nothing leaves the board when it misses.
Week 9, settled
Week 10's market is not broad yet, so this page leads with week 9, as the landing does. What the board said before the games, and what happened. Nothing is removed when it missed.
Leans published, week 9
153 cleared · 108 missed
58.6% of 261 graded cleared, 95% 53%–64%. That interval sits wholly over break-even at −110, 52.4%. 10 leans were void and stay on the ledger. Every lean, graded, on Receipts.
Streaks into week 9
50 carried · 49 broke
101 lines went into week 9 on a streak one of the four rules fired on. 50.5% of the 99 graded carried it, 95% 41%–60%. 2 void. Who carried and who broke.
What a lean is
A gap of 4 points or more between the model and the market.
It is published so it can be graded, not because disagreeing has paid. Every lean stays on the record, cleared or missed.
From the board index · board/nfl/2026/wk10/index.json
Player props
127 of 299 lines shown; 156 carry a lean, at a gap of 4 points or more. Each row sets the market’s chance of the over beside the Second Opinion’s.
124 of the 299 lines are receiving yards, where the model gives no price this week: those rows show the market alone. Every row is complete as shown; nothing on this board is still loading.
Rows 121–127 of 127 · page 4 of 4
Market 52.0%Second Opinion 38.7%
Market 59.1%Second Opinion 34.6%
Chance to clear each threshold
Last 10 games: the share at or above each threshold. Ten games at most: a small sample, drawn as one.
Market 59.1% Second Opinion 34.6%— last ten
The read prices the main line only; the books also listed 3.5, without a price in this file.
This line, game by game
Last 10 against today’s 2.5, not the line posted that week · cleared 6 of 10.
Cleared Missed– not recorded
this season 5 of 8 · career 36 of 72 against 2.5 · 28 of 49 against each game's own closing line (2023-2025 sportsbook closes).
What leans this size have done
Gaps of 8+ points, walk-forward: each lean made with constants fitted only on earlier seasons.
- Leans
- 6,862
- over 852 games
- Cleared
- 53%
- 95% 51%–54%
- Return at the price
- −1%
- 95% −3% to +2%
At this lean’s price (+128) a bet needs 43.9% to break even. The cleared interval sits above it.
wilson_95 · game_block_bootstrap_2000 · research/board_bands.py
Line since open
Opened at 2.5, unchanged.
One read of this line so far (Thu, Nov 12, 8:20 PM ET, 59.1%): no path yet.
| Book | Over | Under | P(over) | Hold |
|---|---|---|---|---|
| fanduel | −172 | +128 | 59.1% | 7.1% |
PIT @ CIN · kickoff Sun, Nov 15, 8:20 PM ET
Market 46.0%Second Opinion 49.7%
Market 59.0%Second Opinion 60.2%
Market 43.1%Second Opinion 30.1%
Market 33.9%Second Opinion 28.9%
Market 51.0%Second Opinion 30.3%
156 leans published for week 10: 27 graded, 129 upcoming, 0 live, 0 void. On this read, 12 cleared, 15 missed.
Receipts: every lean the model has published this season, graded, none removed →
Parlay builder: what a book charges for several of these lines on one ticket →
How the board is read
Market P(over): median of each benchmark book’s de-vigged price at the main line · Second Opinion: baseline-usage-0.4+03571d4710af
- Line
- the main line AT THIS READ: the listed threshold whose median de-vigged over probability across DraftKings, FanDuel and BetMGM is closest to 0.5. is_main is a property of the read, not of the market; it is false only on a row kept for a published lean whose line the main line moved off.
- Gap
- model minus market, probability points, signed
- Lean
- over at gap >= +T, under at gap <= -T
- Longshot
- market P(over) outside [0.15, 0.85]; multiplicative de-vig is biased there - to revisit with Shin or power de-vig once settled data exists
- Void
- market_pulled | inactive | no_snap - a voided lean stays on the ledger
Threshold log (appended, never edited, so every lean is graded against the threshold it was published under): 4 points from 2026-09-01 (decided 2026-09-24 (audit 5.3, unit a-26) before the first board; effective from season start because no board read predates it).
Last read Fri 7:24 PM ET · week 10Break-even at −110 is 52.4%Nothing leaves the board when it misses
Not a sportsbook · no wagers accepted · nothing here is advice · read Fri 7:24 PM ET · 85 reads this week · model baseline-usage-0.4+03571d4710af · board/nfl/2026/wk10