The Carry Trade Hides in Plain Sight

Carry is compensation for holding an exposure whose losses are concentrated in adverse states. Currency carry earns a rate differential while remaining vulnerable to an exchange-rate jump; short-volatility strategies collect option premium while bearing convex losses; financed relative-value trades earn spread while depending on stable funding. The common signature is positive accrual with negative skew. Calm does not prove that the risk has disappeared. Under volatility-sensitive sizing, it can increase the amount of that risk held.

Section 01Why low volatility breeds leverage

One amplification channel is mechanical. Volatility-targeting strategies commonly scale exposure inversely to a forecast of risk, while value-at-risk limits and some margin frameworks become less binding when recent volatility is low. Falling measured risk can therefore permit larger positions even without a stronger expected return. This does not mean every quiet market creates leverage or that added exposure always suppresses volatility. It means the sizing rule is procyclical: the permitted position tends to rise after calm and fall after stress.

The Bank for International Settlements made the point plainly when it dissected the turbulence of August 2024. The vehemence of that episode, it observed, reflected in part a prolonged prior phase of risk-taking amid unusually low volatility, an environment particularly conducive to the build-up of leveraged positions such as currency carry trades and related strategies that profit from contained volatility. The calm had not been empty. It had been filling up with leverage the whole time.

Section 02The calm-to-crowding loop

Figure 8 draws a conditional cycle rather than a universal law. On the build-up half, low realised volatility can relax risk constraints and permit larger carry or volatility-selling positions. Persistent positioning may compress measured volatility in some episodes, but the feedback is neither automatic nor directly observable from volatility alone. The more defensible inference is about fragility: leverage can accumulate across similar trades while the eventual exit capacity remains fixed or deteriorates.

Summer 2025 had the shape of such a stretch. After the tariff episode of the spring, cross-asset volatility subsided; by the Bank for International Settlements' account the VIX, having more than doubled in April, fell back below its pre-announcement level within weeks. For volatility-sensitive strategies, that decline can mechanically permit exposure to be rebuilt. The observation does not establish how much leverage returned, but it identifies the channel through which a calmer risk estimate can alter position size.

Section 03When the loop reverses

August 2024 provides a recent illustration of the reversal. A Bank of Japan rate rise and weak United States data pushed the yen higher, impairing the funding leg of yen-financed carry positions. Leveraged exposures were reduced across currency and equity markets. The Bank for International Settlements estimated foreign-exchange carry positions in the broad region of forty trillion yen, around 250 billion dollars, before the episode and concluded that the unwinding of leveraged trades amplified the initial reaction. Volatility rose sharply before subsiding within days.

The crucial feature is procyclicality. The same volatility-targeting rule that permitted leverage when volatility was low now demands deleveraging as volatility rises: higher measured risk forces position cuts, the cuts move prices, the moves raise measured risk further, and margin increases tighten the screw. The BIS called August 2024 yet another example of volatility exacerbated by procyclical deleveraging and margin increases, and warned that although a broader crisis was avoided, the structural features underpinning such episodes deserve continued attention. That markets stabilised quickly was reassuring, but it was a statement about that episode, not a guarantee about the next one. The loop did not break; it was caught.

volatility-target exposure:   L  ~=  target_vol / forecast_vol

forecast_vol low   ->  permitted exposure rises
forecast_vol jumps ->  exposure must fall  ->  procyclical selling
                      selling can raise measured volatility  ->  cut again

carry P&L:   small + each quiet period,  large - on the jump

The calm had not been empty. It had been filling up with leverage the whole time.

The contribution is not the familiar observation that carry can lose in a shock. It is the interaction between return shape and sizing rule. Negative-skew exposure accrues most smoothly just when backward-looking risk estimates permit the largest position. Fragility is therefore better described by the change in required exposure under a volatility jump than by the current volatility level alone. Carry may be rationally priced; its financing and liquidation schedule determine whether it is resilient.

This research is analysis and commentary for general information. It is not investment advice, an offer, or a solicitation, and it contains no price forecasts. Descriptions of the 2024 episode summarise the cited source; the interpretation is the author's.