Exchange Liquidity Is Not Volume: How to Judge a Venue

Volume is a count of what has already traded. Liquidity is a statement about what could trade now without moving the price. A venue can report enormous volume and hold almost nothing in its book; it can also hold deep resting liquidity while reporting modest turnover.

Traders default to volume because it is the number aggregators display and the number exchanges promote. It is also the number that is easiest to inflate, because reported volume can be generated without consuming any liquidity at all, while depth cannot be faked without capital actually at risk in the book. Published estimates of how much reported crypto volume is fabricated range from roughly half to nearly all, depending on sample and definition.

The gap between the two is not hypothetical. In Kaiko's measurement of global crypto market structure, Binance accounted for 64.3% of global trade volume but 30.7% of global market depth — a factor-of-two divergence between what the venue processed and what it could absorb. Across the top eight venues the figures were closer, at 91.7% of depth against 89.5% of volume, which tells you the divergence is concentrated rather than uniform.

The structural finding in the liquidity research is concentration. Kaiko reports that liquidity has become more concentrated over time, with market makers deploying capital where organic activity already is, and quoting in a tighter range when they do. For ETH specifically, 72% of liquidity sits on five exchanges.

This is self-reinforcing. Market makers go where flow is, flow goes where execution is cheapest, and execution is cheapest where market makers are. The consequence for a trader is that the effective venue set for any given asset is much smaller than the list of exchanges that quote it.

It also means the long tail is thinner than its volume figures suggest. A venue outside the concentrated core that reports competitive volume on a major pair is making a claim that the depth data does not generally support.

Liquidity is not one number. Four measurements together give a usable picture, and each answers a different question about the same book.

• Quoted spread — the distance between best bid and best offer. Answers: what does an instant round trip cost at minimum size?
• Market depth at a band — cumulative resting size within a fixed distance of mid, commonly measured at 0.1%, 1% and 2%. Answers: how much can I move before I move the price?
• Slippage or effective spread — the difference between the price you saw and the volume-weighted price you got. Answers: what did it actually cost?
• Depth resilience — how quickly the book refills after being consumed. Answers: can I do that again in a minute?

Quoted spread is the most visible liquidity metric and the easiest to game. A single small order at the touch produces a tight spread while saying nothing about the size behind it. Two venues can show identical spreads and differ by an order of magnitude in what they will absorb.

The industry consensus on this is clear: execution quality is best measured by realised costs — effective spread, implementation shortfall, realised slippage — rather than quoted spreads alone. Realised slippage should also be segmented by time of day and around event windows, because fill quality degrades exactly when correlated flow arrives, which is exactly when most traders need it.

The practical version for a trader without an execution-analytics stack: measure your own slippage. Record the mid at decision time and the volume-weighted price you received, on every fill, and keep the series per venue and per size band. That series is more informative about your venue choice than any published liquidity ranking, because it is measured at your size.

A defensible venue comparison needs the same pair, the same moment, and the same size. Most published comparisons fail on at least one of the three, which is why they disagree with each other and with your fills.

Sample depth on the same pair across candidate venues at the same instant, at several fixed distances from mid, and repeat across sessions. A single snapshot captures whichever venue happened to have a market maker quoting at that moment, and a venue with a strong Asian-hours book and a thin European one is a different venue depending on when you trade.

• Repeat across sessions and weekdays; report the distribution, not a single reading
• State your sample size and window, so the comparison can be recomputed
• Weight by the size you actually trade, not by the size that makes a venue look best
• Compare volume to depth explicitly; the ratio is the check on the reported figure

Vultax compresses liquidity health into one of the six signal domains inside Vi IQ — market quality, liquidity health, order flow, arbitrage potential, volatility and news sentiment — a 0-100 pre-trade read on whether a market is behaving normally, with the domain scores exposed alongside the headline number so you can see why. How each domain is computed is documented in the methodology.

A composite is a convenience, not a substitute for the underlying measurements, and it is only as good as the book data behind it. Coverage, snapshot cadence and feed reliability vary between venues; we publish those limits rather than assume them away.

Volume tells you what an exchange says happened. Depth tells you what its book will absorb. Slippage tells you what you actually paid. Only the last of these is measured at your size, and it is the one almost nobody records.

If you take one operational habit from this article, make it the slippage log. Everything else here is a way of forming a prior about where to trade; your own fill data is evidence.

The strip at the top of this page is not a screenshot of the day this was written. Every ten minutes Vultax re-reads CoinGecko's exchange tickers for six venues (spread, depth within 2% of mid, anomaly flags), the top fifteen exchanges' reported volume, and five live order books and rewrites the figures; the Vi IQ beneath them scores the same six domains the terminal scores for a pair, computed for BTC on the major spot venues, with any domain that cannot be computed shown as unavailable rather than filled in. Read the numbers as a live check on the argument above, and the revision notes at the end for what has changed since publication.

The pair to compare is the first and the third: dollars resting within 2% of mid across six venues against bitcoin volume reported by the top fifteen venues in a day. The first number is what an order can use; the second is what venues say happened. When the ratio moves it is almost always the volume figure that moved.

Reported volume keeps growing while measured liquidity thins. CoinGecko counts close to $80 trillion of centralised-exchange volume across spot and perpetuals in 2025. In the same period Glassnode measured spot volume at its lowest since November 2023 and CryptoQuant's exchange whale ratio at a ten-month high, a combination analysts read as large holders using thin books as exit liquidity. By July 2026 CoinDesk was describing a survival crisis for smaller venues as day traders disappeared, with BitMEX announcing a September shutdown.

None of that shows in a volume ranking. It shows in depth, in spread under load, and in how often a venue's pairs are flagged as anomalous, which is why the live strip carries those three and not a volume league table. The trust score CoinGecko assigns an exchange still leans on reported volume; the anomaly and stale flags on individual pairs are the part of its data that does not.