Hedge funds & the Treasury basis trade — record leverage in the market that prices everything
Built 2026-06-14. Structured data + edges: macro-treasury-basis-trade.json. Companion to macro-uncovered-risk-pools and macro-money-market-funds.
Leveraged hedge funds are now a dominant marginal buyer of US Treasuries via the cash-futures basis trade — long cash, short futures, financed in repo at near-zero haircuts. The trade is near 2024 record size, top funds run ~18:1, and the March-2020 unwind required ~$1T+ of Fed buying.
1. Size — near record, but not precisely measurable
- Leveraged-fund short Treasury futures near 2024 all-time highs (~$1T+ gross (some measures ~$1.2-1.4T), Fed; vs ~$0.8T in 2019). Press range ~$1T-$2T.
- OFR, verbatim: "Available data do not support precise measurement of the basis trade." Any single "$X trillion" figure must carry this caveat.
- Hedge-fund long Treasury exposure ~$2.3T (Mar 2025, OFR) — part of it the cash leg.
2. Mechanics, leverage, fragility
- Long cash Treasury + short future, cash leg financed via overnight repo rolled to delivery. Three fragility vectors (CFTC/Fed): rate/funding volatility, variation-margin calls, initial-margin hikes — "all salient and pronounced in March 2020."
- Top-10 hedge funds = ~40% of total repo borrowing at ~18:1 (Q3-2024 data, in the Fed FSR, Nov 2025); 73.8% of HF repo at zero/negative haircut (Q4 2022). Macro/RV funds run >6:1 vs the ~2.6:1 industry average — the tail holds the trade.
- HF repo borrowing ~$3.1T (Q2 2025, +154%); Treasury-collateral repo ≥$1.4T.
3. March 2020 — the unwind already broke the market
- 10yr yields surged 54→119bp (Mar 9-18) "in part because of an unwinding by hedge funds of the … basis trade" (OFR); FSB estimates ~$90B force-unwound; HFs sold ~$426B of Treasuries.
- The Fed offered ~$1.5T of repo and bought ~$1T of Treasuries in three weeks — market-function purchases that became QE.
4. Who holds it — the Cayman trail
- Cayman-domiciled hedge funds held ~$1.8T of Treasuries (end-2024) and absorbed $1,225B of net issuance 2022-24 — the second-largest buyer after US domestic, ahead of other foreign private, while the Fed ran off −$1,208B (Fed FEDS Note, Oct 2025).
5. Clearing & warnings
- SEC central-clearing mandate: eligible cash Treasuries clear by Dec 31, 2026, repo by June 30, 2027 (FICC/CME). FICC's Sponsored Service already ~$938B/day. Clearing forces near-zero-haircut bilateral trades into CCP margining — less fire-sale contagion, more CCP concentration.
- OFR 2025: "Hedge fund leverage is near all-time highs." Fed FSR (Nov 2025): basis traders are "an important source of Treasury demand but … relatively sensitive to interest rate volatility and funding market stability." FSB (Jul 2025): 9 NBFI-leverage recommendations.
- Honest tension: the latest OFR/Fed data lean against an "April-2025 unwind" (Treasury-collateral repo held ≥$1.4T, cheapest-to-deliver use rose). The concern is structural (haircuts, ~18:1, $3.1T repo), not a confirmed recent blowup.
Synthesis
The basis trade concentrates record leverage in the $30T Treasury market — the discount rate for the whole AI-capex/credit complex. A rate-volatility shock (which the AI-debt issuance wall feeds through long rates, macro-fdic) forces deleveraging into the same repo/bill plumbing the MMFs dominate (macro-money-market-funds) — a correlated amplifier, not a diversifier.
What is NOT asserted
- No single precise size — the OFR says it can't be measured; figures are position proxies with ranges.
- No confirmed 2025 unwind — latest data point against one; the risk is structural.
- "50-100x" is a trade-level near-zero-haircut figure, not the fund-level (~18:1 top-10) statistic.
- Overlay edges are excluded from the proofs.
Sources: OFR 2025 Annual Report; Fed FEDS — cross-border trail of the basis trade; Fed Financial Stability Report (Nov 2025); FSB — NBFI leverage (Jul 2025); CFTC — basis-trade paper; SEC — clearing compliance extension; DTCC/FICC Treasury clearing.
← Research index · structured data: macro-treasury-basis-trade.json · macro-treasury-basis-trade.md