Research · methodology note · Nov 17, 2026
How the sports flag threshold is set
The scorecard flags a contract when the market price diverges from an independent fair value by enough to be worth recording. For most desks that gate is a flat 6 points. Sports is different: the fair value is a de-vigged sharp-book line, and prediction-market prices track that line very tightly. On the clean record since late August 2026, the typical pregame gap between Kalshi and the sharp book is about 1 to 3 points per sport. A 6-point bar therefore sits far out in the tail, and almost nothing flags: one game in roughly three months. That is not sports being well measured, it is sports being effectively unmeasured.
A threshold set per sport, from the data
So the sports flag threshold is set per sport, from each sport own pregame gap distribution, rather than borrowed from an unrelated global constant. For each sport we take the 90th percentile of every game widest pregame, liquid, underpriced-side gap (one observation per game), then floor it at 3 points and cap it at 6. The floor keeps trivially small gaps from flagging; the cap means this rule can only ever lower the bar for sports, never raise it above the global gate. Every other desk is unchanged.
| Sport | p90 of per-game gap | Flag threshold |
|---|---|---|
| MLB | 1.2pt | 3.0pt (floored) |
| NBA | 2.4pt | 3.0pt (floored) |
| NFL | 2.5pt | 3.0pt (floored) |
| NCAAF | 3.2pt | 3.2pt |
A sport with too little clean data to calibrate keeps the global 6-point gate until it has enough. MLB is the most efficient of the four: even at the 3-point floor it still flags almost nothing, which is the honest result rather than a gap in coverage.
Reproducible, and forward-only
A track record is only trustworthy if the bar that produced it is stable, so the thresholds are frozen, dated rows rather than a figure that drifts with every new game. Recomputing them later adds new rows with a new effective date; it never edits or deletes the old ones, so which rule produced any given flag stays auditable. The change is also forward-only: it sets which new divergences flag from here. It does not reach back and manufacture flags for past games, because the record is append-only and we do not write history after the fact.
The honest caveat
A per-sport threshold has a cost: a sports flag and, say, an energy flag no longer mean a divergence of exactly the same size. We think that is the right trade, because a single global bar measured an efficient market at essentially zero resolution. The gap that triggered each flag is recorded on the flag itself, so the size is always visible, and sports flags are still withheld near settlement and before the binding-artifact cutoff under the sports exclusion rules.
Cite this study
Free to cite with attribution to Scopes. For the citation format, boilerplate, and media contact, see Press & citation.