At 7:40 on a Tuesday morning in March 2023, a portfolio manager in Boston stared at a screen that had turned hostile overnight. A regional bank had failed, another was wobbling, and the corporate bonds she had bought for their steady coupons were suddenly quoted with gaps wide enough to drive a truck through. Nothing about the issuers' factories, pipelines, or balance sheets had changed since Friday. Everything about the price of getting out had.
She did not panic. She sold a third of the position into the first firm bid, kept the highest-quality names, and used the proceeds to buy short Treasury bills yielding more than 4 percent. By the end of the week, her fund had lost less than its peers, and by the end of the quarter, she had cash to buy back the same credits cheaper. The trade that saved her year was not a purchase. It was an exit, executed early, in size, and without apology.
That scene captures a truth the bond market teaches over and over: buying is the easy part. Any desk can accumulate yield when markets are calm. Selling — at the right moment, at an acceptable cost, for a reason the investment committee can defend — is where bond management earns its keep.
A market too large to exit casually
Scale explains why exits deserve a discipline of their own. The U.S. Treasury market alone had roughly $28 trillion outstanding by the end of 2024, according to the Securities Industry and Financial Markets Association, making it the deepest pool of fixed income on earth. Add about $11 trillion in U.S. corporate bonds, more than $4 trillion in municipal bonds, and trillions more in mortgage-backed and agency securities, and the decision to sell is never a small administrative step. It is a transaction in one of the largest markets ever created, and size cuts both ways: Treasuries can absorb enormous flows, while a single corporate issue may trade only a handful of times in a week.
The Federal Reserve's tightening cycle of 2022 and 2023 showed what happens when many holders try to leave at once. The central bank raised its benchmark rate by 5.25 percentage points between March 2022 and July 2023, the fastest tightening in four decades. The Bloomberg U.S. Aggregate Bond Index, the benchmark for most American bond funds, fell about 13 percent in 2022, its worst calendar year on record. Managers who had treated bonds as a quiet ballast discovered, in one brutal year, that duration — sensitivity to interest rates — could overwhelm years of coupon income. A 10-year Treasury yielding under 2 percent at the start of 2022 yielded close to 4 percent by year-end, and prices moved inversely, and violently, all the way.
A manager who bought a 30-year bond at 2 percent and watched the 10-year Treasury touch 5 percent in October 2023 faced a live question: hold a below-market coupon for decades, or sell, realise the loss, and reinvest higher? There is no universally correct answer. There is only arithmetic, mandate, and nerve.
What a good exit decision actually weighs
Serious desks break the decision into four tests, and they run them in order.
First, the thesis test. Why was the bond bought? If it was bought for income, a price fall driven by rates alone may change nothing. If it was bought for a credit upgrade, and the upgrade case has collapsed, the reason to own it is gone even if the coupon still arrives on time. Bill Gross, who built PIMCO into a bond giant on precisely this kind of judgement, often argued that the bond investor's edge lies less in forecasting the economy than in recognising when the original reason for owning an asset no longer holds.
Second, the relative-value test. A bond is never sold in isolation; it is sold against something else. A manager compares the yield, spread, and risk of the holding with the best available replacement, after trading costs. In late 2023, for example, short-term Treasury bills yielded more than many long corporate bonds, a rare inversion of the usual reward for taking duration and credit risk. For some managers, that comparison alone justified exiting longer positions: why accept corporate risk for less income than the government paid for three months?
Third, the liquidity test. Corporate trading remains largely dealer-intermediated, and spreads widen sharply in stress, most for smaller and lower-rated issues. A manager who waits for perfect information may find the bid has vanished. Selling a third early, at a known cost, preserved the ability to sell the rest later; selling everything at once would have advertised distress and invited worse prices.
Fourth, the portfolio test. Does the position still fit the duration target, sector limits, tax position, and clients' need for cash? A bond right for one mandate is wrong for another, and the exit decision belongs to the portfolio, not to the security in isolation.
The counterargument: selling can be the mistake
None of this means selling is always wise. The strongest case against a hair-trigger exit is the record of investors who sold at the worst moment and never bought back. In October 2022, when the Aggregate index was down more than 15 percent from its peak, many retail investors redeemed bond funds in size. Those who stayed were rewarded: the index returned roughly 5.5 percent in 2023 and about 1.3 percent in 2024, and yields that had looked punishing became the foundation for higher future income. A manager who sold a high-quality bond solely because its price had fallen locked in a loss that patience would have softened, and surrendered a coupon stream that, held to maturity, would have repaid principal in full.
Jeffrey Gundlach, founder of DoubleLine Capital, has likewise warned that trading friction erodes returns, particularly in high-yield and emerging-market debt: a desk churning for small yield pickups can end up poorer than a desk that holds.
Neither "always hold" nor "always sell" is a strategy. Holding is a decision, just as selling is, and it must be defended with the same rigour. The manager who holds a deteriorating credit because selling would admit a mistake is not being disciplined. That manager is being stubborn, and the market charges handsomely for stubbornness.
Building an exit discipline before it is needed
The best desks write their exit rules when markets are calm: a loss or spread-widening trigger for mandatory review, a liquidity budget separating what can be sold in a day from what would take a week, pre-approved replacements so a sale does not strand cash, and a post-trade review a month later.
Electronic platforms such as MarketAxess and Tradeweb have changed the mechanics without changing the judgement: portfolio trading lets a manager exit a basket in one transaction, but the fundamental question remains human — at what price is certainty worth more than hope? A manager who can answer, in advance and in writing, why a position would be sold, at what trigger, and into what replacement, has already done most of the work.
The implication reaches beyond trading desks. Wealth advisers guiding retirees through rate cycles, pension trustees matching assets to liabilities, and corporate treasurers managing their own debt all face the same fork: hold or leave, and on what evidence. The firms that treat the exit as a first-class decision — planned, priced, and reviewed — will not avoid every loss. Markets guarantee losses. What those firms will avoid is the worst kind of loss: the one that happened because no one had decided, in advance, when enough was enough.