Deep liquidity and a long coin list pull against each other

Exchange feature lists are written to be compared, which is why they compare badly. They count things that are easy to count — supported coins, order types, API endpoints, advertised yields — and the property that decides what most of those are worth is not on the list, because it cannot be reduced to a number the venue would enjoy publishing.

Liquidity is depth, not volume

Liquidity is the amount you can trade without moving the price against yourself. Concretely: at each price level above the best offer there is a quantity resting on the order book, and if your order is larger than what sits at the top level it consumes the next one, and the next. The average price you end up paying is worse than the price you saw. That gap is slippage, and it is the real cost of size.

So liquidity is not one number. It is a shape, and it is specific to the pair, the venue, the hour, and — crucially — your order size. A market that is deep for a hundred-dollar trade can be thin for a hundred-thousand-dollar one, and the same venue can be excellent in one pair and empty in the next.

Volume is the wrong proxy, and a corruptible one

The number everyone reaches for instead is reported volume, which is a poor substitute on two counts.

It is the wrong measure even when honest: volume is what traded, depth is what could trade now, and a venue can print heavy volume in bursts while holding almost nothing on the book between them.

And it is not reliably honest. Reported volume is self-reported, it feeds rankings that are worth money, and the best-documented public measurement of that incentive found the overwhelming majority of it was not real trading at all. Depth, by contrast, is visible: the order book is published, and you can look at it before you trade rather than trusting a figure about last month.

Why breadth costs depth

“Supports the most coins” and “deepest liquidity” are close to opposites, and the reason is that market making is finite.

Depth exists because someone is willing to quote both sides of a pair and carry the inventory risk of doing so. That is capital and attention, and both are limited. Every additional listing divides them further — so a venue with thousands of markets is, almost by construction, thin in most of them, and a venue with a short list can hold a deep book in each.

Neither is wrong. They are different products, and the feature list presents the first as strictly better because a bigger number looks like more.

The practical version: if you trade two or three major assets, breadth buys you nothing and thinness costs you on every fill. If you want something obscure, the long list is the only place it exists — and you should expect the spread to be where the venue makes its money, exactly as the fee page cannot tell you.

Yields are a credit product wearing a feature’s clothes

“Earn”, “staking rewards” and “interest on your idle balance” appear in feature lists beside order types, as though they were the same kind of thing. They are not. A yield is a payment for someone else using your asset, which means a counterparty, which means credit risk — the question is who is borrowing, against what, and what happens if they cannot return it.

Sometimes there is genuine protocol staking underneath, with its own lock-up and slashing conditions. Sometimes it is a lending market. Sometimes it is the venue’s own balance sheet. These have entirely different failure modes and the same label.

In the EU the question is partly settled by rule rather than by reading: for e-money tokens — the dollar stablecoins most of these products are quoted in — MiCA prohibits paying interest at all, and extends the prohibition to service providers, not just issuers. Where a return is nevertheless on offer, that tells you something about what the product is or where it is offered from.

APIs, and what to check beyond “has one”

For anyone automating, the endpoint list is the least informative part. What decides whether a strategy survives contact with a venue is rate limits and how they are counted, whether there is a websocket feed or only polling, how maintenance windows and partial outages are announced, and whether the sandbox behaves like production. None of that is a feature bullet and all of it is in the documentation, which is worth reading before opening an account rather than after.

Derivatives are a different venue in the same app

Futures and perpetuals share an interface with spot trading and very little else. Positions are leveraged, funding payments move between longs and shorts on a schedule, and a position can be closed by the venue rather than by you when margin runs out. The liquidity that matters is the depth of the derivative’s own book, not the spot pair’s, and the counterparty is the venue’s clearing arrangement rather than another user.

Treating “offers futures” as a feature of a spot exchange is how people end up in a product whose failure mode they have not read.

How to measure depth yourself

The order book is public, so this takes a minute and answers the question the feature list cannot.

Open the book for the pair you care about. Add up the quantity resting within a band you would tolerate — a quarter of a per cent of the price, say. Compare that figure with the size you intend to trade. If your order is a meaningful fraction of it, you are the liquidity, and the price you get will say so.

Do it at the hour you actually trade, not at peak. And do it on each venue you are comparing, in the same pair, within a few minutes — that is a like-for-like comparison, which is more than any table of features will give you.