FanFave
Format sample · not a live forecast
Illustrative slate: 2026-06-19
Engine: v0.1.0 · model hypothetical
Audit ledger: commitment stub — not yet live
Sample format — illustrative figures pending live engine output. No live forecasts have been issued. The factors, weights, bands, and the daily record below are an illustration of the report format only — not a track record. When live publication begins, every figure here is generated by the engine and committed publicly before tip-off.

Indiana @ Oklahoma City

Illustrative game forecast · pre-tip · graded against the closing line
PASS · no edge
Model lean: home side by ~5.0 · Closing line: home −5.5 · Modeled edge vs line: +0.4 pts
Edge does not clear the vig threshold (≈ +1.5 pts required). No pick issued.
Illustrative only. No live forecast pipeline is active.

Read

The model's number lands within half a point of where the market closed. That is the most common outcome by far: the closing line is an efficient, money-weighted estimate, and a domain model that merely re-derives it has found nothing tradeable. The small residual lean toward the home side does not survive the vig.

This sample deliberately shows a pass as its featured game, because passing is what the format does most days. The full reasoning is still published — so you can see the work behind a no-bet, not just behind a bet.

Factor breakdown illustrative · graded vs closing line

FactorWeightBandSituationDissenting case
Rest differentialhome off 2 days, visitor on a back-to-back +0.21 ±0.06 Schedule "Back-to-back fatigue is already the most-priced angle in the market; edge likely double-counted."
Pace / matchup fittransition rate vs half-court defense +0.14 ±0.09 Stable "Pace edge compresses if visitor shortens rotation; sample variance high."
Closing-line valuewhere the market settled at tip −0.29 ±0.05 Market "Line moved a full point toward home post-open; our lean may already be in the price."
Availability (public report)per injury report at line close — no private info +0.08 ±0.10 Availability "Questionable tag unresolved at close; band widened, not narrowed. Do not assume status."
Home-court / travelstandard venue + altitude adjustment +0.10 ±0.04 Stable "No notable dissent; standard adjustment."
Recent form vs schedule strengthopponent-adjusted, last 10 +0.06 ±0.07 Stable "Strength-of-schedule adjustment may overstate home form."
Net edge vs line +0.30 ±0.21 ≈ +0.4 pts after market factor · below the ≈ +1.5 pt vig threshold → PASS
All weights and bands are illustrative. In the live product these are generated from data available to the public before tip-off — not after.

Situation context

  • Schedule spot · visitor on second night of a back-to-backMost-priced public angle · Affects: rest factor, late-game execution assumption
  • Line movement · home −4.5 → −5.5 since openMarket input · Money has moved toward our lean · Affects: closing-line-value factor, whether any edge remains
  • Availability · one rotation player listed questionable at closeUnresolved at line close · Band widened, not narrowed · Model does not assume a status the public did not have

Adversarial vetting excerpt · 1 of 4 challenges

ArchitectProposed: lean home side, ~5.0-point projected margin vs an opening line of −4.5.
Attacker AThe line already moved to −5.5. Your edge was real at open and is gone at close. Grading is against the close, not the open.
Attacker BRest-differential angle is the single most public handicapping input. Assume it is fully priced; do not double-count it.
Attacker CAvailability is unresolved at close. Any implied certainty about who plays is a point-in-time violation. Widen the band.
SynthesizerResidual edge after market and availability adjustments: ≈ +0.4 pts, inside the noise band. Below vig threshold.
Recommender accepted Synthesizer position. Decision node: PASS — no pick issued.

Cases we could not rule out

  • That the questionable player sits, which would move the true number past the closing line and turn this into a live edge. Left unresolved — we do not bet on a status the public did not have at close.
  • That the late line move was sharp money we are on the right side of, in which case passing leaves value on the table. We accept that cost rather than assume our lean is the sharp side.

Why this is a pass, not a pick

  • The modeled margin sits within the closing line's own error band. Re-deriving the market is not an edge.
  • The one angle that pushed our number — rest — is also the most heavily priced public angle. Counting it again is how models fool themselves.
  • Beating the line after the vig requires roughly a point and a half of real edge. We have less than half a point. The honest call is to forecast nothing.