Why Reported Volume Lies: Wash Trading and How to Spot It

Every exchange reports its own volume. No one verifies it. Aggregators publish it, rankings sort by it, projects choose listings by it, and market makers are hired against it — a self-reported number sitting under a large part of the industry's decision-making.

Academic and regulatory studies of unregulated venues have repeatedly found that the majority of reported spot volume on some exchanges does not correspond to genuine economic trading. The exact proportion is contested; that a large share is fabricated is not.

Why anyone bothers

Four incentives, each sufficient on its own. Ranking position on aggregator sites drives user acquisition, and the ranking is sorted by volume. Projects pay listing fees to venues with apparent liquidity. Token teams want their asset to look tradable, which sustains a market for wash trading as a service. And on NFT marketplaces with token incentives, trading against yourself has been directly profitable — the reward for volume exceeded the fee paid to generate it.

The last case is the cleanest illustration, because it needs no deception at all: if a platform pays users to trade, users will trade with themselves until the payment stops.

Fingerprint one: volume that does not fit the book

The strongest single test. Genuine turnover requires liquidity to trade against. If a venue reports enormous daily volume in a pair whose order book shows thin depth within 1% of mid, the two facts are incompatible — you cannot move that size through that book without moving the price far more than it moved.

Compute the ratio: reported 24-hour volume divided by the sum of bids and asks within 2% of mid. Compare across venues for the same pair. Outliers of an order of magnitude are not efficiency.

Fingerprint two: the shape of the trades

Real order flow is messy. Trade sizes cluster around human and algorithmic conventions but scatter widely, and inter-arrival times follow a bursty, heavy-tailed pattern.

Fabricated flow tends to be regular. Round sizes — exactly 1.0, exactly 0.5 — appearing far more often than in a normal distribution. Trades arriving at suspiciously even intervals. A first-digit distribution across trade sizes that departs from what naturally occurring numbers produce. None of these is proof alone; together on one venue they are close to it.

Fingerprint three: volume without price movement

Genuine buying and selling has price impact. If a pair reports substantial turnover with a price that barely moves and a spread that never widens, the trading is not consuming liquidity — which usually means the same party is on both sides.

The inverse also flags: a price that moves violently on tiny genuine volume is a thin market, which is a different problem but equally relevant if you plan to exit.

Fingerprint four: activity that does not match the business

Cross-check reported turnover against independent signals of a real user base — web traffic, app installs, engagement, the number of distinct addresses depositing and withdrawing on-chain. A venue reporting volume comparable to a top-five exchange while attracting a fraction of the traffic is describing users who do not exist.

For on-chain marketplaces this is easier, because settlement is public: look for assets repeatedly sold between a small set of addresses funded from the same source, at rising prices, with no third party ever entering. Wash trading on-chain leaves a permanent record and is the reason NFT volume figures require heavy filtering before use, which the better analytics tools apply explicitly.

What to measure instead

Depth is harder to fake because it is capital at risk: to show real bids, someone must post them and accept being filled. Measure the size available within 0.5% and 2% of mid, at several times of day.

Spread persistence tells you whether the depth is real. A market maker will widen in volatility; a fake book often does not, because nothing is genuinely at risk.

Slippage on a test trade is the honest measure. Execute a realistic size and compare the fill against the quote. This costs a small amount and settles the question completely.

And for custodial venues, verifiable reserves and withdrawal reliability matter more than any activity metric — a venue with genuine volume that cannot process withdrawals has failed at the only thing that counted. That is why our exchange ratings score liquidity from observed depth and spread rather than reported volume, and why the methodology states which figures are self-reported wherever they appear.

The habit to carry

Treat every volume figure as a claim by an interested party until you have checked it against something that costs money to fake. Depth costs money. Traffic costs money. Volume, on a venue that reports its own, costs nothing.