Methodology

What we publish (and what we don't)

We record observed quotes for USDT — a crypto asset — on public peer-to-peer and licensed spot venues, and compare them to freely available reference USD rates. We do not publish, set, or endorse any "parallel exchange rate", and nothing on this site is financial advice.

Data collection

• P2P venue books: full public ad books (price, size limits, posted inventory, payment methods, merchant statistics) are snapshotted every 5 minutes per venue and corridor, and archived unmodified.
• Licensed spot references: public tickers from licensed exchanges (e.g. Luno, VALR) at the same cadence.
• Reference USD rates: a free daily composite (open.er-api.com); central-bank published rates are used where a machine-readable source exists and are labeled as such.

Definitions

• Venue mid: midpoint of the robust best user-buy and best user-sell USDT price on one venue. "Robust best" skips the most aggressive 5% of each price ladder (5th-percentile ask / 95th-percentile bid) because P2P books carry professional bait ads with too-good prices at the top of book; on liquid books this estimator matches licensed-exchange quotes to within ~0.1%. Small books (<10 ads) use a ±5%-of-median trim instead. Medians and depth are computed over the full book, so bait remains visible in the raw data.
• Aggregate mid: median of fresh venue mids for a corridor.
• Parity deviation (the headline series): (aggregate mid − reference USD rate) / reference USD rate, in percent. Per-venue rows use the venue mid instead. This matches the construction used in BIS Working Paper 1340 and IMF crypto-parallel-rate work (WPCPER).
• Depth index: total posted USDT inventory across venues (both sides), and its ratio to the trailing 30-day median. Depth history cannot be reconstructed after the fact — this series only exists because it is recorded live.
• Spread: (best user-buy − best user-sell) / mid, per venue.
• Inferred fills: an ad's posted inventory decreasing at an unchanged price between consecutive snapshots. An estimate, not a trade tape.
• Merchant score (0–10): weighted blend of pricing consistency (0.20), presence during stress periods (0.25), fill patterns vs quick-pulled ads (0.25), venue-reported completion rate (0.20), observed tenure (0.10). Merchants need ≥7 days of observation before scoring.

Staleness

Every number carries its collection timestamp. Data older than 1.5× its collection interval is flagged; older than 3× is warned; sources that stop responding are shown as unavailable, never silently reused.

Downloads

Full-history CSVs are free, without signup: per-corridor rates (/data/NGN.csv), parity (/data/PARITY_NGN.csv), depth (/data/DEPTH_NGN.csv), and the cross-corridor matrix (/data/CROSS_CORRIDOR.csv). Cite as "African Predicter, African Stablecoin Parity Monitor".