A staircase made of other people’s limits

One pair, Bitcoin into USDT on Tron, every provider quoted simultaneously for the same amount, against a spot reference of $84,604.

• 0.01 BTC — 0.47% below spot, 13 providers answering
• 0.05 BTC — 0.43% below, 13 answering
• 0.25 BTC — 0.07% below, 13 answering
• 1 BTC — 0.08% below, 13 answering
• 5 BTC — 0.08% below, 12 answering
• 8 BTC — 0.08% below, 11 answering
• 12 BTC — 0.49% below, 8 answering
• 20 BTC — 0.49% below, 8 answering

It is flat. Then it drops half a percent between eight and twelve. Then it is flat again.

A book does not do that. A book decays continuously, because each additional unit eats a slightly worse price than the one before it. This is a step, and it is in the same place as a different step: three providers stop answering across exactly the same interval.

That is the whole finding. Nothing about the market changed between eight and twelve Bitcoin. What changed is that the providers quoting the best prices have maximums, and above those maximums they simply decline to quote at all.

So the price you see is not a function of size in any smooth sense. It is the best offer from whichever providers will still talk to you at that amount. When the cheapest ones drop out, you fall to the next tier and stay there until the next tier drops out too.

It is a staircase, and the steps are not made of liquidity. They are made of other people’s limits.

We ran the same measurement on Monero to Bitcoin, expecting it to break earlier and harder. Monero is the thinner market, so if anything is going to show liquidity strain it should be that.

• 1 XMR — 0.00713371 BTC per XMR, 8 answering
• 25 XMR — 0.00717322, 7 answering
• 100 XMR — 0.00717649, 7 answering
• 300 XMR — 0.00717526, 7 answering

Up to roughly ninety thousand dollars the rate improves by about 0.6% and then holds. One provider drops out and no step appears.

So the pair we chose because it should have failed first behaved better than the deep, liquid one. Whatever the step on Bitcoin to USDT is, it is not a fact about market depth, because the thinner market does not have it.

The plan was a page explaining when to stop using a comparison site and go and use an order book instead. It would have been an unusual thing for a comparison site to publish, which is exactly why it appealed.

The data will not support it. There is no general rule here. “Large transfers cost more” is true on one pair, for a reason that has nothing to do with being large, and false on the pair where it should have been most true.

What survives is narrower and more honest: on some pairs, above some amount, the providers that quote best stop quoting. The cost of that is a step rather than a slope, and it is specific to a roster rather than to a market.

A ranking built from eight providers is a weaker claim than one built from fourteen, and until this week nothing on the page distinguished them. The rows look identical either way: a list, sorted, best at the top.

The comparison now says how deep it is — how many providers actually priced the amount you asked about — and says it at full depth as well as when it drops. A number that only appears when something is wrong teaches people to ignore it when things are right.

It matters most exactly where it was least visible. At the amount where the quote gets worse, the comparison also gets shallower, and those two things happen in the same step. That is the moment somebody has the most riding on the answer and the least way to tell that the answer is thinner than usual.

We went looking for a cost curve and found a roster artefact. The investigation was worth running anyway, because the thing it refuted was a plausible belief that would otherwise have become a published page — and a page is much harder to withdraw than a hypothesis.

The useful output is not an argument about order books. It is three words on a comparison panel saying how many providers answered, which costs nothing to read and is true every time.