Havoc does not improve opening-gap accuracy after core current-season efficiency is already included.
The Ledger · Team performance
Havoc
Does defensive disruption and offensive protection explain the OV-to-market gap beyond Current-Season Efficiency?
No separate adjustmentTested, but not approved as a standalone Explanatory Factor.Updated August 31, 2026
Decision
Do not display or count a generic Havoc adjustment. Retain havoc as an input for narrower future studies involving quarterback uncertainty, offensive-line disruption, weather, and turnover sustainability.
- Historical seasons
- 2022–25
- Primary sample
- 1,138 eligible games
Start here
What the evidence says
It also fails to improve sampled market movement beyond the core efficiency measures.
The audited result endpoints do not support a standalone havoc adjustment.
Keep havoc for narrower matchup interactions, but adding generic havoc points would double-count information we already use.
2026 prospective implementation
Havoc can be calculated from the existing endpoint, but no generic Havoc Factor is approved because it did not add information beyond efficiency.
Current inputs
- Havoc created and allowed
- Front-seven havoc
- Defensive-back havoc
- Additional source need
- No new provider; future narrow studies require refreshing the existing CollegeFootballData havoc endpoint.
- Refresh cadence
- Only with a versioned narrow study or the in-season efficiency refresh.
- Coverage and limitations
- May support future injury, protection, weather, or turnover-sustainability interactions; it is not a standalone adjustment.
Study results
| Test | Games | Result | Validation |
|---|---|---|---|
| Havoc beyond core · opening | Audited common sample | +0.007 points MAE improvement | Practically zero |
| Havoc beyond core · kickoff | Audited common sample | Worsened accuracy | No incremental gain |
| Havoc beyond core · movement | Audited common sample | Worsened accuracy | No incremental gain |
| Actual-result endpoints | Audited common sample | Worsened accuracy | No ATS or score-correction support |
What we tested
Havoc had a modest standalone relationship with the market gap, but it added no useful information after opponent-adjusted PPA and success rate were included.
- Opponent-adjusted total havoc created and allowed
- Front-seven and defensive-back havoc components
- The same pregame OV Line, provider-level market states, and game controls used for efficiency
Leakage and validation safeguards
- Prior-week-only snapshots with same-week freezing
- Both teams required five completed games
- Leave-one-season-out validation
- Incremental comparison on the same 1,138-game sample
What we learned
- Generic havoc mostly repeats information already contained in PPA and success rate.
- Separating front-seven and defensive-back havoc did not improve the combined model.
- Havoc did not improve paired opening-to-final-recorded movement or actual-margin targets.
What this does not prove
- This result does not imply that pressure or protection never matters in a specific matchup.
- Interactions with quarterback status, offensive-line availability, and weather remain untested.
How uses this finding
Havoc is excluded from point estimates and combined adjustments. When contextually relevant, a matchup may disclose that it was evaluated but not counted separately.
Historical data record
Revalidated August 31 under market-line contract 2026-08-31.2. Havoc is evaluated only after the core efficiency model on endpoint-matched samples. Its opening contribution is effectively zero and it worsens kickoff, movement, and result accuracy.
Audited game-level table: being regenerated under the current market-line contract