Quant Memo
Foundational

Crypto Market Data Pitfalls

The recurring data-quality problems specific to cryptocurrency markets — fragmented venues, wash trading, inconsistent timestamps, and exchange outages — that make naive backtests badly misleading.

Prerequisites: Survivorship Bias

Crypto markets combine several data-quality problems that are each individually present in traditional markets but rarely all at once and this severe. Trading is fragmented across dozens of exchanges with no consolidated tape, so a "price" depends entirely on which venue you pull it from, and prices can genuinely diverge between them for extended periods. Wash trading — an exchange or its users trading with themselves to inflate reported volume — has historically been widespread and hard to filter, meaning a coin's reported liquidity can be dramatically overstated. Timestamps and even historical prices from some vendors have been found to be reconstructed or backfilled inconsistently, and exchanges themselves periodically go offline entirely during volatile periods, creating survivorship-like gaps.

The practical consequence is that a backtest built on a single exchange's historical feed, without checking for wash-trading-inflated volume or cross-checking prices against other venues, can show a strategy as far more liquid and profitable than it would be trading real capital — the historical volume it assumes is executable may be substantially fictional.

A momentum strategy backtested on volume-weighted signals from one exchange might look highly profitable, but if 60% of that exchange's reported volume during the period was later identified as wash trading, the strategy's real achievable size and true historical performance could be a fraction of what the naive backtest implies.

Crypto data pitfalls — exchange fragmentation, wash-traded volume, inconsistent timestamps, and outage-driven gaps — mean a backtest run on a single exchange's raw feed without cross-checking against other venues can badly overstate both liquidity and profitability; always sanity-check volume and price against at least one independent source.

Related concepts

Further reading

  • Alexander & Dakos, 'A Critical Investigation of Cryptocurrency Data and Analysis' (2020)
ShareTwitterLinkedIn