Market Data Quality for Crypto Assets

A crypto quote begins with its venue

Crypto trading spans exchanges, broker interfaces, and decentralized venues with different instruments and conditions. One venue quote is not automatically a consolidated market price. Confirm the pair, quote currency, venue, time, and whether it is a last trade, bid, offer, or index.

Tickers are not unique identifiers. Wrapped assets and unrelated tokens can share symbols. Record the network and contract address where relevant, distinguishing a token from its protocol or application.

Fragmentation affects comparisons

Prices and depth differ with fees, access, settlement routes, and local supply or demand. A gap alone is not an arbitrage opportunity; moving funds may involve delays, limits, fees, and operational risk.

Volume totals depend on coverage and calculation. Providers aggregate spot pairs, derivatives, and venues differently. Keep spot turnover separate from futures volume and check pair conversion, intervals, and included venues.

Treat questionable volume as a research flag

Some researchers have questioned whether reported activity at certain venues represents independent trading. Methods and extent vary; an anomaly alone does not establish misconduct. Explain the provider’s screening method and limits rather than making an unsupported accusation.

Wash trading describes trades arranged to create misleading activity, but public data may not establish it. Unusual turnover or weak price response warrants scrutiny, not a conclusion about intent. Seek transparent methods and corroboration.

Inspect timing, staleness, and transformations

Feeds can lag because of polling, network delays, outages, or provider issues. A retained last value may look live while stale. Check timestamps, gaps, and whether price, volume, and order-book fields share a capture time.

• Preserve source, venue, pair, quote currency, and observation timestamp.
• Flag missing intervals, stale values, outliers, and status interruptions.
• Document conversions, aggregation, exclusions, and index weighting.
• Keep raw observations available when normalized values are displayed.

On-chain metrics depend on definitions

On-chain data follows a network’s transaction model, not a universal measure of economic behavior. An address is not necessarily a person. Transfers may reflect exchange operations, contracts, or user activity; counting them as purchases or adoption needs evidence.

Providers define active addresses, transactions, fees, and adjusted volume differently. Upgrades, token standards, and batching alter interpretation. Name the chain, metric, provider, and time range before comparing networks.

Make the data trail reproducible

Retain raw observations, transformations, timestamps, and exclusions. State whether a figure is venue-reported, aggregated, or ledger-derived; on-chain transparency does not resolve off-chain identity or reporting questions. Data quality belongs to a particular measure and method. This is educational information, not individualized investment, tax, legal, or accounting advice.