The reviewed cases lean away from the affected team, but the interval still includes little or no effect.
The Ledger · Player availability
In-Season Quarterback Uncertainty
When expected quarterback experience or availability differs, does the market move beyond what the team-level rating already captures?
Research candidateThe reviewed cases are promising, but the selected case queue and missing prospective source block a production estimate.Updated August 31, 2026
Decision
Retain experience, confirmed absence, named replacement, and public ambiguity as separate states. A quantified live card requires a sourced pregame state; do not infer a universal injury penalty from the selected review sample.
- Historical seasons
- 2022–25
- Primary sample
- 1,673 in-season FBS-FBS games
Start here
What the evidence says
The sampled movement is promising, but the reviewed case queue is selected and the interval nearly includes zero.
The current evidence cannot support an injury or ambiguity betting rule.
Keep injury, replacement, benching, and ambiguity states separate and require a prospective pregame source before publication.
2026 prospective implementation
Historical states are defined, but broad 2026 detection is not production-ready for injuries, benchings, surprise starters, and unresolved competitions.
Current inputs
- Expected starter state
- Prior-game participation
- Career experience
- Change reason and certainty
- Additional source need
- Requires a reliable prospective news/availability process, likely combining team reporters, public social feeds, and manual verification.
- Refresh cadence
- At least daily during game week and again near kickoff when uncertainty remains.
- Coverage and limitations
- Play-by-play can flag possible ambiguity after a game but cannot establish what bettors knew before the next kickoff.
Study results
| Test | Games | Result | Validation |
|---|---|---|---|
| Expected-starter under-50 rule | 1,673 | 0.013-point final-line MAE improvement | Opening gain 0.009; no movement gain |
| Material experience contrast | Held-out cases | 1.18 to 1.51 points | Fold estimates toward market |
| Reviewed cases · established opening | 42 | 0.60 points away | Interval -0.36 to +1.57 |
| Reviewed cases · opening to kickoff | 24 | 0.71 points away | Interval -0.01 to +1.48 |
| Reviewed cases · kickoff | 24 | 1.48 points away | Interval +0.32 to +2.63 |
| Actual margin vs kickoff | 24 | 2.19 points toward affected team | Interval spans both directions |
What we tested
Automated experience measures add little by themselves. In manually reviewed injury, benching, and ambiguity cases, the audited opening market averages 0.60 points away from the affected team and the selected kickoff cohort averages 1.48 points away. The cases were chosen for review and are not a random population sample.
- Pregame expected starter and career passing attempts
- Prior-game primary passer and play-by-play disruption signals
- Manually reviewed injury, benching, and ambiguity labels
- Independent OnVarsity Line and market consensus
Leakage and validation safeguards
- Later starts do not redefine what was expected before an earlier game
- Expected-starter and realized-primary analyses are separated
- Leave-one-season-out validation
- Manually selected disruption cases are not treated as representative incidence data
What we learned
- A broad uncertainty state may be more observable than a perfect starter prediction.
- Injury-coded cases appear larger than generic experience contrasts.
- Expected starter, ambiguity, benching, and confirmed absence should remain separate retained labels even if some are pooled for estimation.
What this does not prove
- Historical news availability is incomplete.
- The manually reviewed disruption sample was selected for investigation.
- The live 2026 identification process is not yet complete enough for broad publication.
How uses this finding
No generic automatic QB card is created from usage alone. A sourced, prospectively identifiable absence or ambiguity state may display a wide-uncertainty Factor; every state and source remains auditable.
Historical data record
Revalidated August 31 under market-line contract 2026-08-31.2. The review flow contains 50 events, 48 following games, 43 eligible following markets, 42 established-opening observations, and 24 selected kickoff observations. The reviewed queue is not a random incidence sample.
Audited game-level table: being regenerated under the current market-line contract