Methodology
How RipRanks measures card-rip operators, and the rules that keep it honest.
What we publish, and what we don't
We publish two separate things and never blend them. Real value measures money returned, and is only compared across operators sharing an identical settlement basis. The terms— payment methods, cash-out rate, region locks, ID checks, whether the odds are published at all — are stated as facts, sourced from the operator's own site. We deliberately publish no single overall grade: value isn't comparable across operators, and a letter that implied it was would be false precision. Where a fact isn't published, we say “Not stated” rather than estimate it.
Real value
Value = Σ (real odds × each tier's published value-range midpoint), minus a 3–6% haircut calibrated from the pools an operator exposes in full. Where card pools are capped (as on Pullmarket, at 432 top-value cards/tier), card means are inflated and we don't use them. Store-credit returns are always badged as such and never placed in the same column as cash.
Ground truth beats models
Where a live pull feed exists, we sample real reveals (cheap cards included) and prefer that empirical value over the modelled estimate once the sample is large enough. Small samples are labelled “provisional” — we'd rather say “n=12” than fake a precise number.
Provenance on every number
Each figure carries its source (API, sampled, modelled), a date, and — for samples — the size. Stale data freezes an operator and drops it from live rankings. A methodology change is a new version with a dated changelog; we never silently restate old numbers.
Independence
Not affiliated with, or endorsed by, any operator. Every claim is phrased as “advertised X; we measured Y at time T” and backed by a timestamped snapshot. No pay-for-placement, ever.
Per-operator specifics live on each operator's Trust page — e.g. Pullmarket's method.