The Treasury Basis Trade: How Hedge Funds Arbitrage Cash Bonds, Futures and Repo

The Treasury basis trade is a position that buys a Treasury note or bond, finances it in the repo market, and sells Treasury futures against it, so the holder earns the small gap between the price implied by the futures and the cost of borrowing until the futures contract delivers. It is run mostly by relative-value hedge funds, it is large enough to matter for the whole Treasury market, and it works only with heavy leverage.

In short: a basis trader is a warehouse. Asset managers want long Treasury exposure through futures, which ties up little cash. Hedge funds take the other side, short the futures, and hold the actual bonds on borrowed money. If the futures price implies a financing rate (the implied repo rate) higher than the rate the fund actually pays in repo, the fund locks in the difference at delivery. That difference is usually a few tenths of a percentage point a year, so funds borrow 20 to 50 times their capital to make it worthwhile. The trade is hedged against interest rates, but not against margin calls, repo haircuts, or a widening of the gap before delivery. Those are what caused the forced selling in March 2020.

This article builds on how futures prices are set. The delivery mechanics and conversion factors are covered in depth in Treasury futures and cheapest-to-deliver, and the general carry arithmetic in cash-and-carry and the implied repo rate. Here we follow how professionals run the trade, how big it is, and how it breaks.

A common misconception: “the basis trade is a bet on rates”

It is not. The long bond and the short futures move almost one for one with yields, so a parallel move in rates barely changes the position’s value. What the trade is exposed to is the spread between two ways of owning the same Treasury exposure:

Own the bond outright and finance it in repo.
Own a futures contract that delivers the bond later.

In a frictionless market these cost the same. In practice the futures are often a little expensive, because the investors who want them value the low cash requirement. Daniel Barth and R. Jay Kahn describe this in “Hedge Funds and the Treasury Cash-Futures Disconnect” (Office of Financial Research Working Paper 21-01, April 2021; later published as “Hedge funds and the Treasury cash-futures basis trade” in the Journal of Monetary Economics, 2025): a replicating portfolio of Treasury bills and futures had been overvalued relative to notes and bonds, and hedge funds earned the gap by “warehousing” Treasuries for holders of long futures, funding them in repo.

The three legs

Leg 1: the cash bond

The fund buys the security that is cheapest to deliver into the futures contract. CBOT Treasury futures let the short choose which eligible note to deliver, and adjust the invoice by a conversion factor, roughly the price at which each note would yield 6%. The note that costs least to deliver drives the futures price, so that is the one the arbitrage uses.

Leg 2: repo financing

The fund pledges the note to a lender in a repurchase agreement: it sells the note today and agrees to buy it back later at a slightly higher price. The difference is the repo interest. Two features matter:

Haircut. The lender may advance less than the full value. A 2% haircut means the fund funds 2% of the note itself. Barth and Kahn note that Treasury haircuts are often around 3% or less, which implies possible leverage of more than 30 to 1. The OFR’s newer data are starker. Using its non-centrally cleared bilateral repo collection, Ashlyn Cenicola, Robert Mann and Mark Paddrik reported on the OFR blog (August 12, 2025) that 65% of hedge fund repo had zero haircut, and that over 60% of Treasury repo outstanding had a zero haircut.
Tenor. Much basis financing is overnight, rolled every day until delivery. The OFR’s 2025 Annual Report to Congress notes that some hedge funds “use overnight repo and continuously roll over this funding until the Treasury futures delivery date.” That is a rollover risk the fund carries every night.

Leg 3: the short futures

The fund sells futures in a ratio set by the conversion factor, so the delivery obligation matches the bond it holds. Futures need only initial margin, a fraction of notional. The BIS Quarterly Review (September 2023, Box A1, “Margin leverage and vulnerabilities in US Treasury futures”) estimated margin leverage, notional divided by initial margin, at about 70 times for the five-year contract and 50 times for the ten-year, down from about 175 and 120 times before the pandemic.

At delivery, the fund hands over the note, receives the invoice price, and repays the repo. The futures price it sold at, adjusted by the conversion factor, is what it locked in on day one.

How a desk measures the trade

Professionals reduce the position to one comparison: the implied repo rate (IRR) against the actual repo rate.

Gross basis = cash price minus (futures price x conversion factor).
Carry = coupon income minus repo interest over the holding period.
Net basis = gross basis minus carry.
IRR = the return from buying the note now and delivering it into the futures, expressed as an annual money-market rate.

If the IRR exceeds the repo rate, the futures are rich and the arbitrage is to buy the note and sell the futures. A negative net basis says the same thing in price terms. Because the short futures holder owns the delivery options (which note, which day), the net basis is normally slightly positive. The trade is attractive when that option value is being given away.

Worked example: $1 billion of five-year notes

The numbers below are illustrative and were computed in Python. The contract terms are real; the note and prices are hypothetical.

Setup (trade date November 10, 2026, settlement November 11):

Cheapest-to-deliver: a hypothetical 4.125% Treasury note maturing October 31, 2031, priced at 100-08 (100.25).
December 2026 five-year note futures (CBOT, $100,000 face per contract) at 108-22.5 (108.703125).
Conversion factor, using CME’s published formula for the five-year contract (4 years 10 months to maturity from December 1, 2026): 0.9223.
Accrued interest: 0.12535 at settlement; 0.69510 at delivery, which the desk plans for December 31, 2026, 50 days away. No coupon is paid in between.
Repo rate: 3.95%, a few basis points over SOFR, which the New York Fed published at 3.88% for November 9, 2026.

The basis arithmetic:

Measure	Value
Futures x conversion factor	100.25689
Gross basis	-0.0069 points (about -0.22/32)
Carry over 50 days (coupon 0.56975 minus repo 0.55067)	+0.0191 (about +0.61/32)
Net basis	-0.0260 (about -0.83/32)
Invoice price at delivery (futures x CF + accrued)	100.95199
Implied repo rate (act/360)	4.136%
Spread over repo	18.6 basis points

Sizing: $1 billion face costs $1,003.75 million including accrued interest. The hedge is 1,000,000,000 / 100,000 x 0.9223 = 9,223 contracts short. At 70 times margin leverage (the BIS estimate), initial margin is about $14.3 million.

Profit locked in: $1,003.75 million x (4.136% - 3.95%) x 50/360 = $259,700 over 50 days. That is the entire excess return over financing on a billion dollars of bonds.

Return on equity depends on leverage:

Funding structure	Fund equity	Leverage	Annualized excess return on equity
2% repo haircut, no extra buffer	$34.4m	29x	5.4%
Zero haircut, a liquidity buffer that brings leverage to 17.6x	$57.0m	17.6x	3.3%
Zero haircut, futures margin only	$14.3m	70x	13.1%

The 17.6x line is not arbitrary. Barth and Kahn report that in 2019, hedge funds they classified as large basis traders had median balance-sheet leverage of 17.6 and mean leverage of 21, against 1.95 for all hedge funds. Leverage is what turns 18.6 basis points into a return worth having.

Stress test: what a bad week does to the same book

Take the 17.6x fund: $57.0 million of equity, $14.3 million of it posted as futures margin, $42.7 million as a cash buffer. Now apply three shocks together, of the kind seen in March 2020:

Shock	Cash required
The basis widens by 8/32 (cash note cheapens against futures), on $1bn face	$2.5m
The exchange doubles futures initial margin	$14.3m
Repo lenders move the haircut from 0% to 2%	$20.1m
Total	$36.9m, or 86% of the buffer

Three points follow.

The mark-to-market loss is small but out of scale with the profit. The 8/32 widening costs about 9.6 times the $259,700 the trade was meant to earn. It reverses if the fund holds to delivery, but only if the fund can hold to delivery.
The liquidity calls are the real threat. Margin and haircut changes are not losses, yet they consume cash just the same. The 70x version of this fund, with no buffer, could not meet them at all.
The obvious fix makes things worse for everyone. A fund that cannot meet the calls sells notes and buys back futures. If many funds do that at once, the cash note cheapens further, the basis widens further, and the next round of calls is larger. The BIS called this a margin spiral.

Who is on each side: what CFTC data shows

The CFTC’s weekly Traders in Financial Futures (TFF) report splits futures open interest into dealers, asset managers, leveraged funds and others (the report is explained in the commitments of traders report). The table below sums the 2-year, 5-year and 10-year CBOT note contracts at face value ($200,000 per 2-year contract, $100,000 for the others), calculated from the CFTC’s futures-only TFF files.

Report date	Leveraged funds, short	Asset managers, long
February 18, 2020	$659bn	$729bn
March 17, 2020	$584bn	$671bn
March 31, 2020	$535bn	$582bn
January 2, 2024	$920bn	$885bn
December 31, 2024	$1,049bn	$1,092bn
January 6, 2026	$1,212bn	$1,294bn
November 3, 2026	$977bn	$1,328bn

The mirror image is clear: asset managers are long, leveraged funds are short, and the two grew together. The Federal Reserve’s FEDS Note “Recent Developments in Hedge Funds’ Treasury Futures and Repo Positions: is the Basis Trade ‘Back’?” (Barth, Kahn and Robert Mann, August 30, 2023) found that the gap between asset manager longs and hedge fund shorts in these three contracts shrank from $338 billion in June 2022 to $34 billion by May 2023, as hedge fund shorts rose by $411 billion between October 2022 and May 2023. Over the same period, hedge fund sponsored repo borrowing rose from $114 billion to $233 billion.

The proxy overstates the trade

Every Fed paper on this warns that TFF shorts are an upper bound. A leveraged fund can be short futures for reasons that have nothing to do with the basis. In “Quantifying Treasury Cash-Futures Basis Trades” (FEDS Notes, March 8, 2024), Jonathan Glicoes, Benjamin Iorio, Phillip Monin and Lubomir Petrasek combined regulatory Form PF filings with TRACE transaction data. They estimated the trade at between $260 billion and $574 billion at the end of September 2023, against $913 billion of leveraged fund futures shorts in TFF at the time, and concluded that hedge funds had built at least $317 billion of basis-related Treasury holdings since early 2022.

The most recent official estimate is larger. In “Decomposing Hedge Funds’ U.S. Treasury Exposures” (FEDS Notes, June 22, 2026), Phillip Monin estimated the cash-futures basis trade at about $830 billion as of September 2025, roughly double its early-2020 peak and about 3.5% of privately held Treasuries, with the 50 largest funds accounting for about 90% of hedge funds’ gross Treasury exposure. For comparison, Barth and Kahn put the pre-2020 peak at $400 billion to $500 billion.

March 2020: what actually happened

In the second and third weeks of March 2020, the Treasury market, normally the deepest and most liquid in the world, stopped functioning normally. Andreas Schrimpf, Hyun Song Shin and Vladyslav Sushko described the episode in “Leverage and margin spirals in fixed income markets during the Covid-19 crisis” (BIS Bulletin No. 2, April 2, 2020): forced selling of Treasuries “by investors who had attempted to exploit small yield differences” using leverage, in a market where bonds sold off even as risk assets fell.

The evidence on the basis trade specifically:

Futures shorts fell. In the CFTC TFF data above, leveraged fund shorts in the three contracts dropped from $659 billion on February 18 to $535 billion on March 31, and kept falling into April.
Cash Treasuries were sold. Barth and Kahn estimate from Form PF that funds they classify as large basis traders cut long Treasury positions from $756 billion to $652 billion between the ends of February and March 2020, and sold between $91 billion and $105 billion of cash Treasuries as a direct result of shrinking basis positions. Glicoes and co-authors report that hedge funds sold about $173 billion of Treasuries on net during the turmoil, concentrated among likely basis traders.
Margins rose. Barth and Kahn show that CBOT margins track Treasury volatility, and that for the bond contract March 2020 was the highest-margin period in their sample.
The implied repo rate and bill yields separated. The IRR had tracked Treasury bill yields until early March; it then rose while bill yields fell, and the spread peaked around March 17, when the Federal Reserve intervened.

Was the basis trade the cause? Barth and Kahn’s answer is careful: the unwind “was likely a consequence rather than the primary cause” of the stress, which began with sales by other investors, and the Fed’s rapid purchases and repo operations likely prevented a liquidity spiral. The BIS authors put more weight on leverage as an amplifier. Both agree that the structure of the trade, cheap financing rolled daily against margined futures, turns other people’s selling into forced selling.

April 2025: the trade that did not unwind

After the US tariff announcements in early April 2025, the Treasury market sold off sharply and many observers assumed a basis unwind. The OFR’s 2025 Annual Report to Congress found the data inconsistent with that story: hedge fund repo borrowing against Treasuries was at least $1.4 trillion in its bilateral repo collection, with no large declines in April or May, and repo collateralized by cheapest-to-deliver ten-year notes actually increased. Monin’s June 2026 FEDS Note points instead to swap spread trades: about $60 billion of those positions unwound in April 2025 and another $40 billion in May, while his basis estimate rose. The episode is a useful reminder that the basis trade is one leveraged Treasury strategy among several.

The SEC Treasury clearing mandate

The structural response to March 2020 is mandatory central clearing. In December 2023 the SEC adopted rules requiring covered clearing agencies in US Treasuries to make their direct participants clear eligible secondary market transactions, including most repo. On February 25, 2025 the SEC extended the compliance dates by one year, to:

Transactions	Compliance date
Eligible cash Treasury trades	December 31, 2026
Eligible Treasury repo	June 30, 2027

As of September 22, 2026, SEC Commissioner Mark Uyeda said at the US Treasury Market Conference that the SEC “does not currently intend to extend these deadlines.” He also noted that the SEC had registered CME Securities Clearing and ICE Clear Credit as additional Treasury clearing agencies alongside FICC, and that on June 18, 2026 the SEC had granted conditional relief allowing private funds to clear Treasury repo through wholly owned captive subsidiaries.

What this means for the basis trade:

Repo margin. Cleared repo carries clearing-house margin, while much of today’s bilateral repo has zero haircut. The OFR brief “Hedge Fund Participation in Cleared Repo” (Neth Karunamuni and Robert Mann, Brief 26-01, March 3, 2026) cites estimates that about 75% of hedge fund Treasury repo is not centrally cleared. Moving it into clearing should raise the equity the trade requires, which, on the arithmetic above, lowers return on equity.
Visibility and netting. Clearing gives regulators a consolidated view of positions and lets dealers net trades, which frees balance sheet. Karunamuni and Mann find that dealer balance-sheet benefits appear to be a major reason hedge funds clear repo voluntarily.
Concentration moves. Risk shifts from many bilateral relationships to a few clearing houses, which is why their margin models and default resources matter (see central counterparty risk management).

How the market adapts after the December 2026 and June 2027 dates is not yet known.

How professionals run it

Desks that run basis books for a living tend to share a few practices:

Match financing to delivery. Term repo to the delivery date removes nightly rollover risk at the cost of a slightly higher rate. Many funds mix term and overnight funding.
Track the delivery options. The short futures holder chooses what and when to deliver. Desks model how the cheapest-to-deliver note could switch if yields move, because a switch changes the hedge ratio and the value of the position.
Hold a liquidity buffer sized for stress, not for normal days. The stress test above shows why margin and haircut changes, not P&L, set the buffer.
Diversify repo counterparties so one dealer pulling its lines does not force a sale.
Watch crowding. TFF data, sponsored repo volumes and dealer commentary are imperfect but public signals that many funds hold the same position.

What can go wrong

Margin increases. Exchanges raise futures margins when volatility rises (CME’s process is covered in how clearing houses work). That is a cash call on the short leg exactly when markets are stressed.
Repo haircuts and withdrawal. Zero-haircut repo is a lender’s choice, not a right. A lender can widen the haircut, raise the rate, or decline to roll overnight funding.
Basis widening before delivery. The trade shows a loss when the cash note cheapens relative to futures. A fund forced out early realizes it.
Correlated exits. When funds with similar books sell together, they move prices against each other. The Fed and OFR estimates show the trade is concentrated in a small number of large funds.
Delivery and switch risk. If the cheapest-to-deliver changes, the hedged note may no longer be the one the market prices against.
Rule changes. The clearing mandate, and any future changes to bank leverage rules that affect dealers’ repo capacity, change the economics of financing.

The short version

The Treasury basis trade buys the cheapest-to-deliver note, finances it in repo, and shorts the futures, earning the gap between the implied repo rate and the actual repo rate. In our worked example that gap was 18.6 basis points, worth $259,700 on $1 billion over 50 days, which is why the trade is run at leverage of roughly 17 to 70 times. CFTC data show asset managers long and leveraged funds short in matching size, and the Federal Reserve’s latest estimate puts the trade near $830 billion as of September 2025. It is hedged against rates, not against margin calls, repo haircuts and forced selling, which is how it amplified the March 2020 stress. The SEC’s clearing mandate, due for cash trades on December 31, 2026 and repo on June 30, 2027, is the main structural change ahead. This is an explanation of a mechanism, not a strategy to copy.