17 Nov 2026 · 10:44 AM MTUpdated…
Forecast accuracy

Did the predictions earn their confidence?

Keep product completeness, model accuracy and evidence of a betting advantage separate.

What we are testing ↗
Current staking evidence

n=101 · mean -0.22 pts · 95% [-0.56, +0.12] · gate closed (interval includes 0) · ~46 plays for a 0.5 pt half-width · population picks-v0.1 spread, pick in force at week open (114 plays)

Decision chain

“Receipt available” means the instrument returned a measurement. A measured closed gate is still closed.

StageReceipt availableEvidence
0 · the serving contractYes18 declared fields · 272/272 rows complete
1 · our number per gameYes272 of 272 games carry a frozen margin
2 · the market beside itYes163 with an early line · 163 with a close
3 · a decision, before kickoffYespicks-v0.1: 114 spread plays frozen at week open, 0 rows written after kickoff (refused) · legacy ledger: 421 decisions across 12 placement times
4 · settlementYespicks-v0.1: 101 of 114 gate plays graded · legacy ledger: 286 of 421 graded
5 · closing-line valueYespopulation picks-v0.1 · 100 in a quadrant · 0 awaiting a result · 13 with no usable close (0 played with no capture inside 12h)
6 · CLV interval (the staking gate)Yesn=101 · mean -0.22 pts · 95% [-0.56, +0.12] · gate closed (interval includes 0) · ~46 plays for a 0.5 pt half-width · population picks-v0.1 spread, pick in force at week open (114 plays)
6b · CLV vs William Hill close (reference book, not the gate)Yesn=100 · mean -0.22 pts · 95% [-0.56, +0.12] vs William Hill's own close (williamhill_us) · informational; the staking gate is line 6 (market consensus)
6c · CLV, old wager rows (history, not the gate)Yesn=185 · mean +0.01 pts · 95% [-0.28, +0.31] · preseason batch + weekly-v0.1 rows the gate read before 2026-10-03
7 · moneyline centsYespicks-v0.1 straight-up: 101 scored · 13 pending

Decision-chain measurement · Nov 17 · 10:09 AM MT

Exact-score confidence, checked against reality

Games4,897

Archived rolling origins

Most likely score1.6%

Mean probability assigned to modal score

Actual modal hits0.1%

Frequency the most likely score was correct

Diagnostic, not a prospective result

Archived 2008–2025 rolling-origin simulation diagnostic; not a prospective wagering result.

OriginGamesMean modal probabilityActual modal hitsJoint Brier ↓
20082671.8%0.0%1.0056
20092671.6%0.0%1.0058
20102671.6%0.0%1.0060
20112671.7%0.0%1.0060
20122671.7%0.0%1.0063
20132671.6%0.4%1.0058
20142671.5%0.4%1.0065
20152671.6%0.0%1.0064
20162671.6%0.4%1.0063
20172671.6%0.0%1.0068
20182671.6%0.0%1.0067
20192671.6%0.4%1.0065
20202691.6%0.0%1.0066
20212851.6%0.4%1.0065
20222841.5%0.0%1.0066
20232851.7%0.4%1.0069
20242851.6%0.0%1.0068
20252851.5%0.0%1.0067
How to interpret this
Measurement boundary

Historical simulations carry some 2026 rule constants. This is not a fresh leak-free certification.

Measurement boundary

Finite draw counts leave many actual exact scores unobserved; unseen does not establish zero underlying probability.

Measurement boundary

A tighter simulated score distribution is useful only if it scores better against held-out actual outcomes.

Measurement boundary

No staking decision or model promotion is made by this report.

Archived diagnostic run · Sep 23 · 9:27 PM MT

Outcome and price quality

Closing-line value (CLV) compares the recorded price with the close. The source groups zero CLV with its “−CLV” bucket; these labels are reproduced, not corrected in the browser.

Recorded quadrantDecisions
WON +CLV22
WON -CLV30
LOST +CLV12
LOST -CLV36
Decision-level closing-price evidence
GroupEventSelectionTaken lineCloseCLV points
WON +CLVDenver Broncos @ San Francisco 49ersSan Francisco 49ers-2.5-3.00.5
WON +CLVGreen Bay Packers @ Tampa Bay BuccaneersTampa Bay Buccaneers3.53.00.5
WON +CLVCincinnati Bengals @ Miami DolphinsMiami Dolphins7.57.00.5
WON +CLVIndianapolis Colts @ Pittsburgh SteelersPittsburgh Steelers-2.5-3.00.5
WON +CLVMinnesota Vikings @ New Orleans SaintsMinnesota Vikings-1.5-2.51.0
WON +CLVTampa Bay Buccaneers @ Dallas CowboysTampa Bay Buccaneers9.59.00.5
WON +CLVDallas Cowboys @ Green Bay PackersGreen Bay Packers1.51.00.5
WON +CLVIndianapolis Colts @ Minnesota VikingsMinnesota Vikings-5.5-7.01.5
WON +CLVKansas City Chiefs @ Seattle SeahawksSeattle Seahawks-2.5-3.00.5
WON +CLVMiami Dolphins @ New York JetsMiami Dolphins4.53.51.0
WON +CLVAtlanta Falcons @ Tampa Bay BuccaneersTampa Bay Buccaneers-1.5-2.00.5
WON +CLVCarolina Panthers @ Green Bay PackersGreen Bay Packers-3.0-3.50.5
WON +CLVIndianapolis Colts @ Jacksonville JaguarsJacksonville Jaguars-7.5-10.02.5
WON +CLVKansas City Chiefs @ Denver BroncosDenver Broncos1.50.01.5
WON +CLVDallas Cowboys @ Indianapolis ColtsIndianapolis Colts5.54.01.5
WON +CLVJacksonville Jaguars @ Tennessee TitansJacksonville Jaguars-9.5-10.51.0
WON +CLVLos Angeles Chargers @ Baltimore RavensLos Angeles Chargers5.54.51.0
WON +CLVLos Angeles Rams @ Arizona CardinalsLos Angeles Rams-7.5-10.02.5
WON +CLVMiami Dolphins @ Indianapolis ColtsIndianapolis Colts-1.5-4.53.0
WON +CLVMinnesota Vikings @ Green Bay PackersMinnesota Vikings3.02.750.25
WON +CLVPittsburgh Steelers @ Cincinnati BengalsCincinnati Bengals-3.5-4.00.5
WON +CLVSeattle Seahawks @ Las Vegas RaidersSeattle Seahawks-3.5-6.53.0
WON -CLVArizona Cardinals @ New York GiantsNew York Giants2.52.50.0
WON -CLVDetroit Lions @ Carolina PanthersCarolina Panthers3.54.0-0.5
WON -CLVJacksonville Jaguars @ Cincinnati BengalsJacksonville Jaguars2.52.50.0
WON -CLVKansas City Chiefs @ Las Vegas RaidersLas Vegas Raiders4.54.50.0
WON -CLVNew England Patriots @ Buffalo BillsNew England Patriots7.07.00.0
WON -CLVNew York Jets @ Chicago BearsChicago Bears-3.5-3.50.0
WON -CLVBaltimore Ravens @ Atlanta FalconsBaltimore Ravens-4.53.25-7.75
WON -CLVBuffalo Bills @ Los Angeles RamsBuffalo Bills2.53.5-1.0
WON -CLVChicago Bears @ Green Bay PackersGreen Bay Packers1.51.50.0
WON -CLVDetroit Lions @ Arizona CardinalsDetroit Lions-5.5-5.50.0
WON -CLVHouston Texans @ Tennessee TitansHouston Texans-6.5-6.50.0
WON -CLVLas Vegas Raiders @ New England PatriotsNew England Patriots-3.5-3.50.0
WON -CLVNew York Giants @ Washington CommandersNew York Giants3.53.50.0
WON -CLVChicago Bears @ Atlanta FalconsChicago Bears-1.5-1.0-0.5
WON -CLVLos Angeles Rams @ Las Vegas RaidersLos Angeles Rams-7.0-3.25-3.75
WON -CLVNew York Giants @ Houston TexansNew York Giants7.510.25-2.75
WON -CLVTampa Bay Buccaneers @ Carolina PanthersTampa Bay Buccaneers2.53.0-0.5
WON -CLVArizona Cardinals @ Dallas CowboysArizona Cardinals6.57.5-1.0
WON -CLVBaltimore Ravens @ Buffalo BillsBuffalo Bills-3.5-3.0-0.5
WON -CLVCleveland Browns @ Pittsburgh SteelersPittsburgh Steelers-3.5-3.50.0
WON -CLVLos Angeles Chargers @ Los Angeles RamsLos Angeles Rams-7.0-6.5-0.5
WON -CLVArizona Cardinals @ Seattle SeahawksSeattle Seahawks-11.5-8.5-3.0
WON -CLVCincinnati Bengals @ Atlanta FalconsCincinnati Bengals-3.5-3.0-0.5
WON -CLVLos Angeles Rams @ Washington CommandersLos Angeles Rams-5.5-4.5-1.0
WON -CLVLas Vegas Raiders @ San Francisco 49ersSan Francisco 49ers-9.5-9.0-0.5
WON -CLVNew York Jets @ Kansas City ChiefsKansas City Chiefs-10.0-9.5-0.5
WON -CLVBuffalo Bills @ New York JetsBuffalo Bills-7.5-7.50.0
WON -CLVKansas City Chiefs @ Atlanta FalconsKansas City Chiefs-3.5-3.0-0.5
WON -CLVNew England Patriots @ Detroit LionsDetroit Lions-2.5-2.50.0
WON -CLVSan Francisco 49ers @ Dallas CowboysSan Francisco 49ers-4.0-4.00.0
LOST +CLVPhiladelphia Eagles @ Jacksonville JaguarsJacksonville Jaguars-3.5-8.04.5
LOST +CLVArizona Cardinals @ Los Angeles RamsLos Angeles Rams-12.0-12.50.5
LOST +CLVNew Orleans Saints @ New York GiantsNew York Giants1.51.00.5
LOST +CLVPittsburgh Steelers @ Tampa Bay BuccaneersTampa Bay Buccaneers2.51.750.75
LOST +CLVDenver Broncos @ Arizona CardinalsArizona Cardinals5.04.50.5
LOST +CLVGreen Bay Packers @ Detroit LionsDetroit Lions-3.5-4.00.5
LOST +CLVChicago Bears @ Seattle SeahawksSeattle Seahawks-3.0-3.50.5
LOST +CLVLas Vegas Raiders @ New York JetsNew York Jets1.5-1.53.0
LOST +CLVTennessee Titans @ Cincinnati BengalsCincinnati Bengals-10.0-13.53.5
LOST +CLVCleveland Browns @ New Orleans SaintsNew Orleans Saints-2.5-3.51.0
LOST +CLVDetroit Lions @ Miami DolphinsMiami Dolphins7.06.50.5
LOST +CLVGreen Bay Packers @ New England PatriotsNew England Patriots-2.5-3.00.5
LOST -CLVAtlanta Falcons @ New Orleans SaintsNew Orleans Saints-2.5-1.0-1.5
LOST -CLVDallas Cowboys @ Houston TexansHouston Texans-3.0-3.00.0
LOST -CLVIndianapolis Colts @ Washington CommandersWashington Commanders3.55.0-1.5
LOST -CLVLos Angeles Rams @ Philadelphia EaglesPhiladelphia Eagles3.53.50.0
LOST -CLVMiami Dolphins @ Minnesota VikingsMinnesota Vikings-10.5-10.0-0.5
LOST -CLVTennessee Titans @ Baltimore RavensBaltimore Ravens-11.5-11.50.0
LOST -CLVCleveland Browns @ New York JetsCleveland Browns1.52.5-1.0
LOST -CLVDenver Broncos @ Los Angeles ChargersLos Angeles Chargers3.53.50.0
LOST -CLVSan Francisco 49ers @ Seattle SeahawksSeattle Seahawks-3.0-3.00.0
LOST -CLVBaltimore Ravens @ Cleveland BrownsBaltimore Ravens-4.5-3.0-1.5
LOST -CLVBuffalo Bills @ Las Vegas RaidersBuffalo Bills-6.0-5.0-1.0
LOST -CLVCarolina Panthers @ Philadelphia EaglesPhiladelphia Eagles-1.53.0-4.5
LOST -CLVHouston Texans @ Jacksonville JaguarsJacksonville Jaguars-3.0-2.5-0.5
LOST -CLVLos Angeles Chargers @ Kansas City ChiefsLos Angeles Chargers7.57.50.0
LOST -CLVNew York Jets @ New England PatriotsNew England Patriots-7.5-7.0-0.5
LOST -CLVSeattle Seahawks @ Denver BroncosSeattle Seahawks-2.5-2.50.0
LOST -CLVTennessee Titans @ Indianapolis ColtsIndianapolis Colts-6.5-6.50.0
LOST -CLVWashington Commanders @ San Francisco 49ersSan Francisco 49ers-7.5-7.0-0.5
LOST -CLVCincinnati Bengals @ Baltimore RavensBaltimore Ravens-3.53.0-6.5
LOST -CLVCleveland Browns @ Tennessee TitansTennessee Titans1.51.75-0.25
LOST -CLVDallas Cowboys @ Philadelphia EaglesPhiladelphia Eagles-1.04.0-5.0
LOST -CLVNew England Patriots @ Chicago BearsNew England Patriots1.53.0-1.5
LOST -CLVPittsburgh Steelers @ New Orleans SaintsNew Orleans Saints2.03.0-1.0
LOST -CLVSan Francisco 49ers @ Atlanta FalconsSan Francisco 49ers-4.5-2.5-2.0
LOST -CLVMinnesota Vikings @ Detroit LionsMinnesota Vikings2.53.5-1.0
LOST -CLVNew England Patriots @ Miami DolphinsNew England Patriots-7.5-7.0-0.5
LOST -CLVPhiladelphia Eagles @ Washington CommandersPhiladelphia Eagles3.55.5-2.0
LOST -CLVBuffalo Bills @ Minnesota VikingsMinnesota Vikings3.03.5-0.5
LOST -CLVDenver Broncos @ Carolina PanthersCarolina Panthers2.53.0-0.5
LOST -CLVHouston Texans @ Los Angeles ChargersLos Angeles Chargers3.55.0-1.5
LOST -CLVJacksonville Jaguars @ Baltimore RavensBaltimore Ravens7.59.5-2.0
LOST -CLVNew York Giants @ Philadelphia EaglesPhiladelphia Eagles2.53.0-0.5
LOST -CLVTampa Bay Buccaneers @ Chicago BearsChicago Bears-6.0-6.00.0
LOST -CLVCarolina Panthers @ New Orleans SaintsNew Orleans Saints1.01.00.0
LOST -CLVHouston Texans @ Cleveland BrownsHouston Texans-4.5-3.5-1.0
LOST -CLVWashington Commanders @ New York GiantsNew York Giants-1.5-1.50.0

How candidates earn promotion

TestRequired comparison
Forecast qualityProper scores on held-out game and player distributions, plus calibration and error slices
Horizon valueIdentical scored population at each immutable pregame lock, with price available at that time
Mechanism valueAblation and shuffled-feature control; pooled uncertainty by game and season
Decision valueClosing-line value, settlement, price/execution costs and correlated exposure
Serving integrityThe scored model/data versions are the versions actually published

Build commitments

Source: executable build plan. A completed item is not proof of betting edge.

ItemStateWork
E-01doneE-01 — Rolling origin 2004→2025: fit ≤ season s−2, calibrate on s−1, simulate every game of s from its
E-08doneE-08 — Weekly prospective issuance at the TUE lock, frozen beside v1 and the rival, graded the followin
E-06doneE-06 — Plant an effect of known size in the play data and require every gate to find it. Also the place
E-07doneE-07 — The smallest detectable effect per layer, per window, game-clustered. It replaces the single 0.1
E-09doneE-09 — Run `world-model-compiler-v1/cross-team-response-experiment.json`, which is `DESIGNED_NOT_RUN`:
J-10doneJ-10 — Delete the stale "irreducible" text; fail the build on `irreducible\|noise floor\|ceiling` appli
J-07doneJ-07 — Force `ORDER BY` on every sampled frame and hash the frame. This removes the ~0.006 pts/wk row-o
J-06doneJ-06 — Score and fit on the unconditional population, with `E[pts] = P(play)·E[pts \| play]` and DNPs a
J-05doneJ-05 — The durable results store, `data/results.duckdb`. It exposes `record(script, git_sha, population
J-02doneJ-02 — Remove the `HOME_FIELD_MARGIN/2` double count (1.748 pts/game). Home field is derived from sched
J-01doneJ-01 — Remove `mkt_total`, `mkt_spread`, `mkt_team_total` from `team_volume` and refit. At serve time t
E-02doneE-02 — CRPS (margin, total, team totals, player lines), log loss (winner), energy score (joint: both QB
J-03doneJ-03 — Re-run every headline accuracy number on clean features and the full population; publish before/
J-09doneJ-09 — Cross-fit every calibration, isotonic and stack weight that is in-sample today (D-232: 0.9984 in
E-05doneE-05 — The Residual Attribution Ledger of §1. `oracle_pairs.py` tests candidate sensors **in pairs**, b
E-04doneE-04 — One score per layer on the full issued population. The layers are availability, action, personne
J-08doneJ-08 — Mutation harness: inject a known defect per test and require a red result. 73 of 156 tests curre
E-10doneE-10 — Re-run what the old path measured wrongly (the queue follows this table).
E-03doneE-03 — Frozen-release backtest RED on 544 held-out games
J-11doneJ-11 — One full trace: reproduce a sealed issuance bit for bit

Desk roster

8 staffed, 0 partial, 1 struck. Every job the operating model charters for this desk (docs/WHAT-I-WAS-MISSING.md #4), against what actually runs in this repo -- not what the tab above implies.

JobStatusEvidence
Chief Forecast Accuracy Officer
Owns prospective forecast quality across season, week, game, team, player, joint, and live horizons.
staffedbin/chain-report.py and buildsite/render.py's accuracy() page already report measured forecast quality end to end.
Season & Futures Forecaster
Forecasts the evolving distribution of team and league outcomes across the season.
staffedresearch/c20/playoff_mc.py Monte Carlos the season's evolving playoff distribution.
Weekly Slate Forecaster
Owns accuracy across the complete weekly slate and its common shocks.
staffedplayerweek/forecast.py builds the complete weekly slate every cycle.
Game Outcome Forecaster
Owns each game’s joint distribution, from winner through score and game shape.
staffedbin/decide-week.py and playerweek/sim.py's game rollout own the joint spread/total/moneyline distribution.
Team State Forecaster
Forecasts the team-level quantities that generate games and player opportunity.
staffedthe team_week table, read by sim_world.py and forecast.py, carries the team-level quantities (total_line, plays_l3, team_spread) that generate games.
Player Opportunity Forecaster
Owns who plays and how much opportunity each player receives.
staffedresearch/playgrain/participation_impute.py and playprob.py forecast who plays and how much, though participation_impute.py's own 0.85 Jaccard gate is not yet met.
Player Efficiency & Stat Forecaster
Turns opportunity into distributions for rates, box-score stats, fantasy points, and props.
staffedplayerweek/forecast.py and playerweek/sim.py turn opportunity into rate, box-score, and fantasy-point distributions.
Joint Distribution & Reconciliation Scientist
Makes player, team, game, week, and season forecasts coherent without erasing disagreement.
staffedplayerweek/simplayers2.py's correlation-matrix gate (F-08) makes player, team, and game draws coherent -- it passes both halves only since its correlation criterion was recalibrated from raw sign to BH-significant sign (5 of 21 cells still disagree in sign, none significant), and sim.simulate() serves RB/WR/TE intervals only -- QB stays on the fitted curve. [independent review 2026-09-23]
Live / In-Game Forecaster
Updates all remaining outcome distributions as plays reveal state and policy.
struckno live win-probability or in-game update path exists; every forecast here is pre-kickoff.