Crypto Market Data API: Coverage, History, and Validation

Define coverage before comparing endpoints

Count instruments by venue and market type rather than quoting one large symbol total. Spot, perpetual futures, dated futures, options, indexes, and tokenized instruments have different identifiers, trading rules, and historical start dates.

A production catalog should preserve the venue symbol, base and quote assets, status, precision, contract type, settlement asset, listing time when available, and every observed status change.

• Name every venue and market type.
• Separate active, pending, suspended, and delisted instruments.
• Publish first and last available timestamps by stream.
• Do not describe a single venue as consolidated crypto coverage.

Validate time, sequence, and gaps

Crypto trades continuously, so missing intervals cannot be explained by a weekend calendar. Track source event time, receive time, sequence identifiers, reconnects, and expected intervals. A zero-volume candle and an uncollected candle are different states.

Backfills and live streams should converge into one canonical schema while retaining their source runs. Test duplicate trades, out-of-order updates, book-sequence gaps, late corrections, and UTC interval boundaries.

Check operational and legal independence

Reliable serving should come from owned storage and normalized tables, not a live dependency on the venue during each customer request. Raw payloads, retrieval manifests, checksums, and deterministic replay make source outages recoverable.

Technical access does not automatically grant redistribution. Rights must be classified per stream and use case before records become public. DataCedar currently keeps the independent Binance capture behind an internal-validation boundary while those serving rights are verified.

How DataCedar preserves the evidence

DataCedar separates acquisition from serving. Permitted source responses are retained with retrieval time and identifiers, normalized into DataCedar-owned tables, checked against expected coverage, and exposed through a stable versioned API. A collector can be replaced without changing the customer contract or making an upstream provider a runtime dependency.

Every research stream carries effective and known-at time where the distinction matters. Rights-restricted, unavailable, partial, stale, and genuinely empty states remain visible, so a backtest can fail closed and a buyer can see the product boundary before committing engineering time.