After both teams have five prior games, current efficiency helps explain part of the opening OV-to-market gap.
The Ledger · Team performance
Current-Season Efficiency
Do opponent-adjusted changes in PPA and success rate explain part of the gap between the OnVarsity Line and market consensus?
SupportedApproved as a market explanation after both teams have five prior games.Updated August 31, 2026
Decision
Use as a matchup-level Explanatory Factor beginning only after both teams have five completed games. Cap the displayed estimate at 2.5 points and present it in half-point increments.
- Historical seasons
- 2022–25
- Primary sample
- 1,163 eligible games
Start here
What the evidence says
The signal adds modest information after open; most of its value is in explaining the market level, not forecasting a dramatic move.
The Factor improves market explanation, not prediction of the final score against the market line.
Use it after the five-game threshold, cap the estimate, and describe it as a market explanation rather than a pick.
2026 prospective implementation
The rule and production threshold are defined, but live 2026 cards begin only after both teams have five completed games and the in-season efficiency feed is refreshed.
Current inputs
- Opponent-adjusted PPA
- Opponent-adjusted success rate
- Pregame team baselines
- Independent OV Line
- Additional source need
- No new provider. Refresh the existing CollegeFootballData play/advanced-stat inputs.
- Refresh cadence
- Weekly after games are final; never include the current or same-week games.
- Coverage and limitations
- Not eligible before both teams have five completed games. The current matchup example is a development-only mockup.
Study results
| Test | Games | Result | Validation |
|---|---|---|---|
| Established opening vs OV Line | 1,133 | 3.21% MAE improvement | Improved in 3 of 4 held-out seasons |
| Kickoff market vs OV Line | 679 | 4.62% MAE improvement | Improved in 3 of 4 held-out seasons |
| Opening-to-kickoff movement | 669 | 1.21% MAE improvement | Improved in 4 of 4 held-out seasons |
| Actual margin vs OV Line | 1,163 | 0.26% MAE improvement | Small and not an ATS result |
| Actual margin vs market endpoints | 679–1,133 | Worsened MAE | Opening -0.36%; kickoff -0.50% |
What we tested
The combined Factor improved held-out final-recorded market-gap accuracy in every tested season once both teams had played at least five games. A smaller relationship was already visible at the opener, and paired-provider movement provided modest additional support.
- Pregame opponent-adjusted PPA change from each team’s entering profile
- Pregame opponent-adjusted success-rate change from each team’s entering profile
- Focal-game-excluded OV Line plus provider-level opening and final recorded CFBD lines
Leakage and validation safeguards
- Only games from earlier weeks enter each snapshot
- The focal game and all same-week games are excluded
- Leave-one-season-out validation across four seasons
- Early observations are shrunk with 150 prior pseudo-plays
- The conclusion held under 75-, 150-, and 300-play priors
- Opening-to-final movement uses only providers with both market states
What we learned
- The useful signal begins after five completed games for both teams.
- The final-recorded market-gap result was stable across every held-out season and prior-weight sensitivity.
- The stronger final-recorded relationship than opening relationship is consistent with current-season information being incorporated after opening.
- The Factor did not improve actual margin against either market endpoint, so it remains a market explanation rather than a score correction.
What this does not prove
- The Factor explains market differences more strongly than final outcomes.
- PPA and success rate are correlated and are intentionally presented as one combined Factor.
- The estimate may overlap with player availability or other information observed by the market.
How uses this finding
Eligible matchups may display Current-Season Efficiency as a quantified Factor. Its estimate contributes to the rough combined Factor estimate and is labeled as aligned with or counter to the market.
Historical data record
Revalidated August 31 under market-line contract 2026-08-31.2. The five-game rule leaves 1,133 established-opening cases, 679 kickoff-observed cases, 669 paired movement cases, and 1,163 completed games against the OV Line.
Audited game-level table: being regenerated under the current market-line contract