There is a ceiling over every market-making book, and nobody who sells returns likes to talk about it.
A market maker earns a few basis points on every dollar of flow it intermediates. Earnings scale with turnover - and turnover has a hard, observable limit, because you cannot be a meaningful fraction of a market that is not there. A desk claiming to run fifty million dollars a day through a pool that itself trades one million a day is not describing a strategy. It is describing a fiction. We hold ourselves to a written rule about this: our flow in any pair must stay a modest fraction of what that venue itself actually trades, and the rule is enforced by the same plausibility checks that police the rest of our numbers.
The belief this piece is here to break: that returns are made by the strategy alone. They are made by the strategy times the market it stands in. And markets do not sit still - liquidity migrates between venues, and lately between forms of the same asset. The book has to be sized against where it is, not where it was.
The map changed
This spring we measured, venue by venue, where the trading in our assets actually happens. Not where it is convenient for an on-chain fund to say it happens - where it happens.
Read the bottom row first, next to the row above it. WBTC - the older wrapped form of Bitcoin - now trades about eighty-four thousand dollars a day in its main on-chain pool, while cbBTC, the newer form, carries thirty-six million: Bitcoin's on-chain liquidity did not die, it changed wrappers. And both are rounding errors next to the four point nine billion dollars of Bitcoin the combined order books of six centralized exchanges turn over in the same twenty-four hours. The pool trades in a day what the books trade every ninety seconds. Nothing disappeared. It relocated.
The other rows tell the same story in milder dialects: Ether's pools carry around twelve million a day against nearly six billion on the books; even SOL, the healthiest on-chain market we trade, is outweighed several times over by its own centralized books.
The edge - the spread between what impatient flow pays and what patient capital earns - did not disappear. It never does. It is simply being paid out at a different address.
Why staying put would become dishonest
Here is the part that matters more than any routing decision, because it is an accounting statement, not a technology one.
On-chain pools are this book's home, and for most of our assets they remain a genuinely large habitat - SOL's pools alone turn over nine figures a day. But liquidity rotates, and a pool that thins underneath a book flips from home to ceiling - and then to something worse. Keep growing a book inside a pool the flow has left, and there is a day when your reported turnover becomes a material share of that venue's entire volume. Past that point you are no longer harvesting a market's spread. You are the market - pressing on your own prices, filling against your own shadow, and reporting numbers that no longer describe an edge at all.
There are only two exits from that squeeze. You can keep sizing against the venue liquidity has left - report flow it could never have hosted, and hope nobody divides your numbers by its - or you can keep the arithmetic and follow the flow. The first is how track records end up fiction. We took the second.
What changed
Execution now treats a measured set of six centralized venues as first-class destinations alongside the on-chain pools. A controlled share of the book's flow executes against those deeper books instead of queuing against thin ones - and for Bitcoin, whose liquidity has moved furthest from its old on-chain address, nearly all of it does.
No pathos in this. Depth, spreads and turnover on each venue are sampled continuously from the venues' own books and tape; the routing engine sizes against what was observed in the last hours, never against a brochure number; a venue whose data goes stale silently drops out of the set. Execution follows liquidity. That is the whole design.
What did not change
The ledger is the same ledger. Every unit of flow, on-chain or off, still has to fit inside the same modest fraction of its venue's own measured turnover - deeper venues widen the denominator, not the rule. Attribution still splits what the strategy earned from what prices did. And the self-policing got stricter, not looser: the plausibility checks grew a new criterion that tests the realized edge on centralized-venue flow against those venues' own measured spread economics, and flags anything the venue could not plausibly have paid. A wider road, with a speed limit that is now actually enforced on it.
You will see the effect in the reported figures over the coming weeks - the same strategy, standing in deeper markets, does more work per day. This piece exists so that the explanation is on the record before the numbers move, not after. That ordering is the whole point of how we report.
The trade-off
Deeper liquidity is not a free upgrade, and it would be against everything this journal stands for to present it as one.
Centralized venues bring risks the pools never had: counterparty exposure to the venue itself, outages at exactly the moments liquidity matters most, settlement and custody mechanics that are operational rather than atomic, a market structure that is less publicly inspectable than a pool contract, and a basis between on-chain wrappers and native books that can widen when it is least convenient. Routing itself is a new surface for error - a misread book or a stale feed is a way to lose money that simply did not exist before.
We accept those risks selectively, not wholesale - only where the measured capacity gain is large enough to justify them, which is precisely why Bitcoin flow moved almost entirely and SOL flow moved least. Each of these risks now lives in the same inventory of things that can go wrong as everything else we do, with the same discipline: named, bounded, and priced - not waved away.
The point
Execution is not loyalty to a venue.
It is loyalty to the market that can actually carry the flow - measured, not remembered.
When liquidity moves and a book keeps sizing against where it used to be, that is not conviction - it is bad accounting. And good investing starts with good accounting.