Rates algos hinge on liquidity plumbing
J.P. Morgan says the spread of automated treasury execution lags other asset classes because rates algorithms depend on where and how liquidity can be sourced, not just better models. The piece argues that execution quality in rates is now as much about venue integration, ECN access and connectivity as it is about internal algos (globaltrading.net).
A stock algorithm can usually hunt in one pond. A Treasury algorithm has to fish across several ponds at once, and J.P. Morgan’s Brooke Bauer says that is why automated rates trading has spread more slowly than in foreign exchange. (globaltrading.net) In United States government bonds, the hard part is often not deciding whether to buy or sell a 10-year note. The hard part is finding the best pool of liquidity at that exact second, on the right venue, with the right counterparty access and the right connection speed. (globaltrading.net) That market is split into at least two big neighborhoods. One is interdealer trading, where dealers and principal trading firms trade heavily on electronic platforms, and the other is dealer-to-customer trading, where asset managers, hedge funds, and other clients face banks through different protocols. (bankofcanada.ca) A principal trading firm is a fast electronic market-maker that trades its own capital rather than client money. Federal Reserve research says these firms became major players in electronic interdealer Treasury trading years ago, especially on automated venues. (federalreserve.gov) That sounds like perfect territory for algorithms, but the plumbing is uneven. Some Treasury liquidity sits in a central limit order book, which is an electronic queue where the best displayed bids and offers meet, while other liquidity sits in request-for-quote systems, where a client asks selected dealers for prices. (newyorkfed.org) If your algorithm only sees one of those pipes, it is like using a maps app that knows highways but not side streets. It may produce a clean-looking execution report while missing better prices or larger size somewhere else. (globaltrading.net) This is why Bauer’s point is less about smarter math and more about market access. J.P. Morgan says rates execution quality now depends on venue integration, electronic communications network access, and connectivity alongside the bank’s internal logic. (globaltrading.net) Regulators and market groups have been pushing on the same bottleneck from another angle. The Treasury Market Practices Group updated best-practice recommendations on price transparency across trading platforms in February 2024, which shows how much attention is now on what traders can actually see across venues. (newyorkfed.org) The pressure is rising because the Treasury market keeps getting more electronic, not less. Coalition Greenwich said daily Treasury volumes started 2025 up 8% from a year earlier, even as volatility was down 14%, which means firms are trying to automate into a market that is still structurally fragmented. (greenwich.com) There is also a resilience angle. New York Federal Reserve researchers wrote that broader all-to-all trading could help in stress periods by letting more participants trade directly with each other when traditional intermediaries are constrained. (newyorkfed.org) So the race in rates is shifting from “who has the best algorithm” to “who has the best wiring.” In Treasury trading, an algorithm without broad liquidity access is like a race car with one lane open. (globaltrading.net)
