Validated, with
the lower bound shown.
Every figure here is HVP-compliant: walk-forward out-of-sample, multi-season, with the Beta-Binomial 95% CI lower bound shown alongside the headline number. Where a sport has no validated commercial tier yet, we say so plainly.
What's
next.
Sports where we have signal but not yet a tier that clears the HVP gate. Listed publicly because saying so is the standard.
In calibration. No spread or moneyline tier has cleared our validation standard yet, so there is no betting number to publish here. We list it rather than leaving it off the page.
In calibration. No betting tier has cleared our validation standard yet, so there is nothing to publish here. A deeper data history is the next step before a regular-season number can land.
In calibration. No betting tier has cleared our validation standard yet, so there is nothing to publish here. When one does, it moves onto the tables above.
In calibration alongside WNBA. Same status: no validated betting tier yet, nothing to publish until one clears the standard.
How we
measure.
Every figure on this page is HVP-compliant: walk-forward out-of-sample across multiple independent test seasons, with Beta-Binomial 95% CI lower bounds shown alongside each headline number. Closing-line baselines are the mathematical break-even rates at standard pricing, not measured market outcomes. Where a sport has no validated commercial tier yet, we say so explicitly rather than publishing inflated headline figures.
Read the full methodology →Questions,
answered.
What does HVP-compliant mean?
HVP is VAR's internal Honest Validation Protocol. Every win-rate or ROI figure must be measured walk-forward across at least three independent test seasons (with a documented variance buffer when fewer are available), reported with the lower bound of a Beta-Binomial 95% credible interval, and audited against the production code path. Numbers that fail those rules cannot be cited in marketing, dashboards, or this leaderboard.
Why is the CI lower bound shown alongside the headline number?
An 80% win rate on n=20 sounds great until you compute the confidence interval and realize the lower bound is around 60%. The point estimate is for marketing. The lower bound is the number to plan around. Showing both is honest; showing only the point estimate is the failure mode HVP was written to prevent.
What does 'in calibration' mean?
The model is live and producing predictions, but no betting tier has met our validation standard yet, so there is no headline number to show. We list these sports here rather than leaving them off the page. When a tier clears the standard, it moves up onto the leaderboard with the rest.
How often will this leaderboard update?
Updated when material new validation results land, not on a fixed marketing cadence. HVP rule one requires at least three independent test seasons before a number changes meaningfully, and large samples accumulate over weeks or months, not hours. The 'last updated' date at the top of the page reflects the most recent HVP audit cycle.
Want this edge
in your stack?
VAR builds simulation infrastructure for franchises, media companies, and prediction-market operators. Plug into the same models that power this leaderboard.
See solutions