The most informative print in global macro this week did not cross a crypto tape. It landed on the sterling funding desks of a handful of London banks that have worked out how to borrow from the Bank of England below the policy rate and put the proceeds to work in bonds. Cheap central bank money, converted into duration, carrying a positive spread with essentially no credit risk and very little basis risk.
On paper, that is free money.
That is also the entire content of the story as it reached me. Banks in London are capitalizing on a bond strategy fueled by cheap Bank of England financing. No facility name. No drawdown size. No participant list. No maturity profile of the assets purchased. No leverage multiple. No hedging disclosure.
My first instinct is not to trade it. It is to ask what the trade is short. Every carry position is a short position in something — usually volatility, frequently liquidity, occasionally the credibility of the institution supplying the funding. If you cannot name the thing you are short, you are not running a strategy. You are being paid to warehouse a tail.
Context: The Funding Window and What It Was Built To Do
The Bank of England has run this playbook before, and the template tells you almost everything about the incentives now in play. The Term Funding Scheme with additional incentives for SMEs, launched in 2020, offered banks four-year funding priced close to Bank Rate, with an extra allowance tied to lending to small and medium enterprises. The design was a subsidy with a policy condition bolted to it. The condition was credit. The intended transmission path was mechanical and unglamorous: cheap funding, cheaper loans, real economy activity, higher nominal GDP. Drawings across that program and its predecessor ran into the hundreds of billions of pounds.
What this story describes is the same cheap funding pointed at a different destination. Bonds.
The carry mechanics are simple enough to fit in one cell of a spreadsheet. Borrow at the facility rate. Buy gilts or high-grade sterling paper. Hedge duration with swaps or futures. Book the spread. If the funding rate sits below the point on the curve where you buy, the trade pays you to exist. Roll the funding. Collect the difference. Mark the collateral. Repeat.
There is no innovation in that. There is no proprietary edge in that. There is a policy rate that has been set below the market clearing price of term funding for institutional balance sheets, and a small number of desks positioned to arbitrage the difference. The value of the trade is not intellectual. It is positional.
Which is precisely why it deserves a second read. The carry trade is not the story. The carry trade is the symptom. The story is that the transmission channel broke, and the liquidity the central bank pushed into the banking system did not emerge on the other side as credit. It emerged as leverage against the government's own curve.
Now the part that matters for anyone who prices risk rather than narrates it. A carry trade funded at a floating rate and invested at a fixed one is a short volatility position wearing a bond's clothing. You are comfortable in a low realized-vol regime, because the funding leg reprices gently and the collateral marks are quiet. You are insolvent in a high realized-vol regime, because the funding reprices immediately while the asset leg does not. You are not long bonds. You are short the path of rates, and you are short it with borrowed money.
And sterling has a recent, thoroughly documented rehearsal of what that costs.
End of September 2022. A fiscal event, a violent gilt selloff, thirty-year yields moving in excess of a hundred basis points across a handful of sessions. The leveraged gilt structures inside pension LDI portfolios took margin calls on collateral whose value had just collapsed, and the only way to meet those calls was to sell the collateral. Forced selling into a market where the marginal buyer had withdrawn. The Bank of England had to open a temporary purchase facility to break the loop. The realized loss was not the lesson. The mechanics were the lesson. When the asset you post as collateral is the same asset you are levered long, a price decline triggers a margin call that forces you to sell the thing whose price already fell.
The loop is self-reinforcing. It runs at the speed of the collateral mark, not at the speed of anyone's judgment, and not at the speed of any committee's meeting schedule.
That is why a headline about London banks and cheap central bank financing should not be filed under banks doing bank things. It should be filed under collateral architecture. And it should be filed again, in a different drawer, under the crypto basis trade, because they are the same structure with different plumbing.
Core: Order Flow, Destination, and the Crypto Mirror
I trace a trade the way I once traced a contract line by line. Origin. Intermediary. Destination. Hedge. Exit. Failure mode.
Origin: the central bank balance sheet, accessed through a facility window priced below market term funding. Intermediary: a London bank trading book with the balance sheet capacity and the regulatory reporting apparatus to hold duration. Destination: gilt or high-grade bond inventory. Hedge: interest rate swaps or futures, stripping out the directional exposure and leaving the spread. Exit: roll the funding at the same price, unwind the hedge, keep the carry. Failure mode: the funding does not roll at the same price, or the collateral marks down far enough that the margin call forces the unwind.
One structural consequence of this flow is routinely missed by the QT-is-tightening crowd. If the central bank is running down its asset holdings while simultaneously offering banks subsidized funding to buy the same class of assets, the two operations partially offset each other. The balance sheet shrinks in name while the duration stays parked in the banking system, financed below market. That is not easing in the conventional sense and it is not tightening either. It is a change in the identity of the holder while the price support persists. Anyone modeling the policy stance off the headline size of the balance sheet alone is modeling the wrong variable.
The second consequence is reflexivity, and it is the one that should worry a policy committee. The same institution that supplies the cheap funding also authors the volatility that determines whether the trade survives. When the central bank's stated objective is to suppress inflation by tightening financial conditions, and its operational reality includes a subsidized funding channel that supports leveraged duration, the two signals are in direct contradiction. Markets are good at resolving contradictions. They resolve them by discounting the credibility of the signal that is cheapest to abandon.
I have traded this structure from the other side, and the uncomfortable part is how normal it feels while it is working. In 2020 I built a delta-neutral structure on Uniswap V2, sold volatility against stablecoin pairs, and stayed flat through an August correction that took roughly forty percent off the equity of people chasing yield with leverage. I did not win that month because I was smarter about direction. I won it because I had pre-committed to the shape of the position, not the level of the price. The same discipline applied in the gilt carry trade gives you the same conclusion: the spread is not the risk. The funding is the risk.
In 2022, after the Terra collapse, I moved from centralized derivatives onto on-chain perpetuals and worked the spread between centralized and decentralized price feeds. That was the same trade again — a positive-carry structure whose real exposure was to the reliability of the funding venue, not to the direction of the asset. And in 2024 I structured a box spread against the gap between spot Bitcoin ETFs and the trust that preceded them, locking a modest risk-free return on a large notional across three time zones. That was a carry trade. Funding leg. Asset leg. Spread. Roll risk. Operational risk. The London gilt carry and my ETF box spread share the same skeleton. The only genuine differences are the venue and the collateral.
Lay the three structures side by side and the identity is unavoidable. The LDI portfolio was long duration financed short. The London bank carry trade is long duration financed short. The crypto cash-and-carry basis is long spot financed short through perpetuals or futures. All three are short volatility. All three are short liquidity. All three are long the continued good behavior of a funding venue they do not control. What separates a boring trade from a blown account is never the width of the spread. It is the reliability of the funding and the correlation between the collateral and the position.
There is a crypto-native leg to this that most desks have not connected to the gilt market at all, and it is the one I would put on the watch list first. The stablecoin float is a leveraged long of the front end of the dollar curve. Issuers hold short-dated government paper and repo. The yield on that float is the revenue line. Supply expands when the carry is attractive to the distributors moving it. Which means stablecoin supply is not a sentiment indicator and it is not a crypto adoption indicator. It is a money market indicator. When global dollar funding tightens, the float mechanics tighten with it, and the crypto market loses its settlement layer at precisely the moment it needs it most. That is a plumbing risk sitting inside every basis book on every venue.
Here is where the crypto version of the trade is structurally better and structurally worse than the London one, and the distinction deserves precision. Better, because the collateral is publicly auditable. Positions, funding rates, open interest, liquidation levels, insurance fund balances — all of it is legible to anyone who builds the right query. In the LDI episode, the market found out the size of the problem only after the problem forced itself into view. In crypto, the leverage is on the ledger before the break. Worse, because there is no lender of last resort. No temporary purchase facility. No committee that can announce a backstop over a weekend. Twenty-four-hour settlement, continuous margin, auto-deleveraging, and an insurance fund that is a rounding error against a genuine cascade.
Read those two facts together and you get the actual insight. The crypto basis trade is a real-time instrument for pricing the exact reflexive risk that the gilt carry trade conceals inside a quarterly mark. The loop that took days in sterling can complete in minutes in crypto. And the option market knows it, even when the spot market does not. If London banks are structurally short rate volatility, financed by the institution that sets the policy rate, the convexity does not sit in the bond. It sits in the wings. Time decays options; patience decays noise. The premium is not in the carry. It is in the tail that the carry is paying you to ignore.
Liquidity dries up; logic remains solvent. That sentence has made more money for me than any directional call I have ever made, and it applies verbatim here.
Contrarian: The Consensus Read Is Pointed the Wrong Way
The consensus read is comfortable and wrong in two directions at once. Cheap central bank funding, the argument goes, means liquidity, and liquidity means risk assets up. That conclusion depends on a destination assumption nobody has verified. Liquidity injected into the bond market is not liquidity available to risk assets. It compresses term premia. It suppresses realized volatility. It raises the price of everything and funds nothing. That is the difference between a liquidity wave and a credit wave, and only the second one generates earnings. Anyone marking their crypto book up because a sterling funding window is open is importing a balance-sheet fact into a risk-asset thesis that does not follow from it.
The second error is accounting. To the bank running the carry, the trade looks close to risk-free in the P&L. To the central bank, it is a contingent liability against institutional credibility. The moment the market understands the funding window as a subsidy for leveraged duration, the window becomes a policy problem. No institution keeps underwriting a trade that amplifies the very volatility it is mandated to control. The reversal does not require a crisis. It requires a meeting.
Now the part I expect to be least popular with the people reading this in a bull market. The reflexive line that crypto is the hedge against banking stress is a misreading of liquidity mechanics dressed up as a safe-haven thesis. When the regional bank complex cracked in March 2023 and Bitcoin rallied, the move was not a flight to safety. It was a repricing of the expected policy path plus a stablecoin depeg that forced a scramble into the asset with the cleanest collateral and the only settlement layer that never closed. The asset that rallied was not the hedge. It was the venue. Confusing the two is how a portfolio ends up long exactly the wrong thing during exactly the right crisis.
And the deepest blind spot sits on the crypto side, in retail-facing yield products. The basis trade is marketed as income. It is a short-vol product with a tail that is invisible because funding rates print continuously and are formatted to look like a coupon. Structure survives where sentiment collapses. The person who gets out of a collateral loop is not the person with the best thesis. It is the person who can see who is levered and to whom. Audit trails are the only true alpha in chaos, and in a market where the collateral is public, the audit is available to anyone willing to do the work instead of the narrative.
Takeaway: The Watch List and the Question That Matters
The trade is not unanalyzable, even from a headline with no numbers attached. It is simply under-specified, and under-specified trades have a specific defense: watch the funding, not the spread.
Sterling front end first. The spread between the facility rate and SONIA or term OIS is the subsidy, expressed as a single number. If that spread widens, the subsidy is deepening and the carry is being encouraged. If it narrows, the trade is being unwound, voluntarily or otherwise. Thirty-year gilt yield and realized volatility come second, with the 2022 range as the reference frame for what a broken loop looks like on a chart, rather than a narrative.
Third, the offsetting-operations test. Compare the pace of balance sheet reduction against the size of the funding window still outstanding. If duration is being shed from the central bank's book while being reabsorbed by subsidized bank balance sheets, the net policy stance is far looser than the headline suggests, and the dislocation between the announced stance and the operational reality is the trade.
On the crypto side, one number earns its place at the top of the screen: annualized perpetual funding against the risk-free rate. When perp funding falls below what a T-bill pays, the basis carry turns negative, and negative carry on a leveraged book has exactly one resolution. Watch the CME futures basis curve for the shape of that compression, and watch stablecoin float as a money market variable rather than a sentiment variable.
We do not predict the wave; we engineer the board. The question is not whether the London gilt carry trade makes money. The question is whether the funding window survives the first time the market prices it as a subsidy rather than a service. And whether your crypto book has an answer ready for the day someone asks the same question about the venue holding its collateral. The ledger remembers what the market forgets.