The spread was 0.00154 bps. The cost was 5.19.
And a zero maker fee does not rescue it. We checked, because that is the first thing anyone assumes.
Market making sounds like the safe way to earn in crypto: quote both sides, capture the spread, stay flat. We measured what that actually pays on BTCUSDT.
| spread captured per round trip | 0.00154 bps |
| cost per round trip | 5.19 bps |
| measured maker advantage | 1.83 bps |
The obvious objection is fees. So we asked what happens at a zero maker fee — the tier the largest firms actually trade on. It still loses. The gap is not a fee problem.
The finding that outlived the strategy
The useful result came out of the wreckage. Adverse selection — how much the market moves against you right after you are filled — is brutal at short horizons and decays by roughly a hundredfold between sixty seconds and one day, to 0.1301 bps.
We had predicted around 1 bps and were wrong by a factor of nine, in the direction that matters: longer horizons are far more economically survivable than the short-horizon numbers suggest. That single measurement reopened a line of work the market-making result had closed.
Why publish a failure
Because the arithmetic is checkable and the conclusion is not obvious. A ratio of 3,400 to one is not a near miss to be optimised away — it says the strategy is unavailable at this latency and this fee tier, and no amount of tuning changes that.