Methodology

Every score and figure on OpenMoney is produced by versioned, published methods. Each method page is generated from the code that implements it, so the page cannot drift from what runs.

• Evidence — what counts as a source, how freshness and sample size set confidence, and how link checks downgrade stale sources.
• Scorecards (scoring v1) — six dimensions (Pay, Fees, Payout, Stability, Transparency, Scam risk), their inputs, the weight table, and how missing evidence marks a composite as partial. Available when the registry launches.
• Scam signatures — the rule codes, the evidence each rule requires, and the fixture set used to measure precision.
• Passport analytics — net hourly, fee rate, volatility, tax set-aside (an estimate, not tax advice) and their formulas.
• Matching — hard filters, expected-value scoring from reported data, and the fairness tests run on every release.
• Settlement — AP2 mandate verification, spend-policy enforcement, the bounty lifecycle state machine, the routing fee, and the no-custody ledger invariant.
• Rankings — how the Open Opportunity Rankings group and rank opportunities by kind and region on each scorecard dimension and the composite, and how movers are computed against the closest prior quarter.
• Aggregates — the k-anonymity threshold (n ≥ 20 per cell) and suppression rules.

Changes to any method are recorded in the changelog with the version they took effect.