Eurodollar to SOFR: the 26.161bp migration

Four decades as the benchmark

For four decades, the Eurodollar future was the world's benchmark short-rate contract. CME launched it on 9 December 1981 as the first cash-settled futures contract ever traded. It was quoted on the same "100 minus rate" convention still used for the SOFR future today, and it settled in cash against three-month US dollar LIBOR at expiry. By 2004 it was trading more than a million contracts a day, and it held that position, the most actively traded interest rate future in the world, for the next two decades.

A rate with no future

Then LIBOR itself was retired. After the LIBOR manipulation scandal, regulators decided to replace it with SOFR, a rate built on actual, secured overnight Treasury-repo transactions rather than banks' self-reported borrowing estimates. That decision left a practical problem with no real historical precedent: millions of live Eurodollar contracts, some expiring years into the future, referenced a rate that was about to disappear. Cancelling these contracts outright was not an option, and asking every counterparty to renegotiate bilaterally was never realistic for a market this size.

A single, engineered conversion

CME's answer was a single, engineered conversion. On 14 April 2023, it converted essentially every remaining open Eurodollar futures and options position, contracts expiring after June 2023, into the economically equivalent SOFR futures and options contract. It did this by adding a fixed spread of 26.161 basis points to each Eurodollar settlement price. Only the May and June 2023 contracts were left alone, allowed to trade out to their own natural expiry. About 7.5 million contracts moved across in this one step. A second, related conversion followed on 3 July 2023, moving cleared zero-coupon and remaining LIBOR swaps, alongside roughly US$4 trillion notional of cleared USD LIBOR swaps converted earlier in the process, onto SOFR terms.

A market that barely moved

The most striking part of the story is how little actually seemed to happen. CME's own SOFR futures and options open interest rose by only about 66,000 contracts once the dust settled, even though 7.5 million contracts had just changed hands overnight. The reason is netting: most traders already held some SOFR exposure, so the converted Eurodollar position mostly cancelled against a position they already had. By the end of 2023, SOFR futures and options open interest stood at 48 million contracts, and average daily volume that year, nearly 6 million contracts, was 34% above the highest annual average Eurodollar futures and options had ever reached across their entire 42-year history.

Each side of a Eurodollar position was doing something specific. A trader short Eurodollar futures was locking in a borrowing rate, protected if rates rose. A trader long was locking in a lending or investment rate, protected if rates fell. The difference between the two reference rates is what makes the conversion non-trivial: SOFR is a secured, repo-based rate, while LIBOR was an unsecured, interbank-borrowing rate, so SOFR sits structurally below LIBOR by a credit and liquidity premium. Simply swapping "LIBOR" for "SOFR" in a contract, with no adjustment, would have quietly shifted value from one side of every position to the other. Adding the fixed 26.161bp spread to the settlement price is designed to cancel out exactly that structural gap, so that on the conversion date itself, nobody's economic position changed. Only the reference rate and the label did.

The episode turns the course's cost-of-carry and no-arbitrage logic into a market-design problem rather than a pricing problem. The spread here is not discovered by trading; it is fixed once, by a regulatory and industry convention (the ISDA and Bloomberg five-year historical median), and then applied identically to every contract, with no negotiation and no case-by-case pricing. It also illustrates how a benchmark rate works more generally: a benchmark is a convention the market has agreed to use, and when the convention changes, every contract that referenced the old one needs an explicit, engineered bridge to the new one. The near-invisible change in open interest after the fact, 66,000 contracts of net movement against 7.5 million gross conversions, is the clearest evidence of this. That gap illustrates how a well-designed, exchange-cleared migration can retire a benchmark that trillions of dollars of exposure depended on without a market disruption, precisely because the spread was fixed in advance and applied mechanically.