Evidence before assertion

Intelligence methodology and public evidence standard

BettorsACE is designed as a sports-intelligence and decision-support platform. Production model outputs are review controlled, evidence linked, and separated from wagering execution, staking, custody, settlement, and payout authority.

Canonical production boundary: BettorsACE may analyze markets and publish approved intelligence. It does not place or settle wagers, accept bankroll deposits, custody stakes, or initiate payouts.

How the intelligence lifecycle works

1. Source stateOdds, event context, injuries, weather, historical evidence, and source freshness are captured before analysis.

2. Model stateModel version, inputs, probabilities, fair-price estimates, confidence, and applicable ensemble context are attached to the decision record.

3. Review gateAutomated outputs remain shadow/review controlled. A model score does not itself grant publication or transaction authority.

4. Evidence ledgerApproved public records can be evaluated against closing prices and outcomes so performance claims remain inspectable.

What BettorsACE measures

Win rate alone is not a sufficient model-quality measure. BettorsACE is structured to evaluate calibration, market-relative price quality, closing-line value, confidence-bucket behavior, sample size, model version, source freshness, and final result. This reduces the incentive to cherry-pick isolated wins.

Calibration: whether stated probabilities behave like observed frequencies over meaningful samples.
Closing-line value: whether published prices compare favorably with later market prices.
Provenance: which evidence and model state produced a released intelligence record.
No-action discipline: passes and insufficient-evidence states remain valid outputs.

Public evidence interfaces

The production backend maintains public-ledger interfaces for released picks, aggregate performance, and individual evidence records. These interfaces are intended to support product UI, audits, research, and future machine-readable proof surfaces without granting transaction authority.

/v1/ace-picks — released public intelligence records.
/v1/ace-picks/performance — aggregate public performance evidence.
/v1/ace-pick — individual public record lookup.

Commercial boundary

BettorsACE can charge for software subscriptions, intelligence reports, creator/operator workflows, and approved API access. Those products sell analytics and software services. Bankroll funding, wager execution, settlement, staking/custody, and payout processing are outside the canonical production product.

Claims discipline

No model eliminates risk or guarantees profit. Public claims should be supported by observable production evidence and bounded to the time period, model version, sport, market, and sample represented by that evidence. If reliable evidence is unavailable, the correct output is to say so rather than invent a prediction or performance statistic.