EM FX Carry with Crash Hedges

Thesis (edge)

Some countries pay much higher interest rates than others. If you borrow in a low rate currency like the dollar or the yen and hold a high rate currency like the Mexican peso or the Brazilian real, you earn the difference simply for holding the position. In a textbook world that difference would be exactly cancelled out by the high rate currency weakening. In practice it usually is not. High yielders tend to hold their value or even strengthen for long stretches, and the carry is pocketed.

The catch is that the payoff pattern looks like selling insurance. You collect a steady stream of small gains, and then occasionally you hand a large chunk of it back all at once. The old desk saying is that carry goes up by the stairs and down by the elevator. This version of the trade accepts that reality instead of pretending it away. It runs the carry basket, and it permanently spends a slice of the income on protection so that the elevator ride is survivable rather than terminal.

The edge, if it exists, is not that the crash never happens. It is that the premium collected across many quiet months exceeds the cost of the hedge plus the losses in the bad months.

Where it works (regimes)

The strategy earns its keep when global risk appetite is stable, dollar funding is easy, commodity prices are firm and volatility is low. In those conditions capital flows toward the high yielders, which supports the currency on top of the interest income, and the trade earns from both directions at once.

It suffers when the dollar rallies hard, when global funding tightens, when a commodity shock hits an exporter, or when any single country stumbles into a political or fiscal crisis. Notice that all of these arrive with very little warning and they tend to arrive together, because the same investors are crowded into the same handful of currencies and they all reach for the exit on the same day.

Assume the strategy is long global risk appetite in disguise. It is not a diversifier from equities. In the worst equity months it very often loses at the same time.

Signals

• Carry: take the difference between the forward rate and the spot rate, annualized. Use the forward points rather than the headline policy rate, because the forward already embeds what it actually costs you to fund and roll the position. A currency can look high yielding on paper while the forward market has already priced in the depreciation.
• Volatility filter: divide carry by recent realized volatility. A currency paying 12 percent that swings 25 percent a year is a worse deal than one paying 6 percent that swings 8 percent a year.
• Fundamental screen: exclude countries showing the classic pre-devaluation pattern. Falling foreign exchange reserves, a widening current account deficit financed by short term money, and inflation running away from the policy rate. High carry from a country in that state is not a premium, it is a warning.
• Trend confirmation: if a currency is already in a sustained downtrend against the dollar, do not buy it just because it pays well. Falling knives pay very high carry on the way down.

Portfolio construction

Rank the surviving currencies by volatility-adjusted carry. Go long a basket of the top names, funded short in dollars, or split the funding across the dollar and a genuinely low yielding currency to avoid a pure dollar bet.

Weight positions by inverse volatility rather than equally, so that a wild currency does not quietly dominate the risk. Cap any single currency at a hard limit, because concentration is how carry books die. Rebalance monthly. Higher frequency rebalancing just pays more spread without materially improving the signal.

The hedge is a permanent line item, not a tactical decision. The simplest version is to hold a standing position in out-of-the-money options that pay off when the dollar rises sharply, sized so the premium consumes a fixed fraction of the expected carry income each year. Buying protection only when you feel nervous does not work, because protection is most expensive exactly when everyone else feels nervous too.

Risk model

The main risk is not day to day volatility. It is the jump. A managed currency can trade in a narrow band for two years and then move 25 percent in a single session when the peg breaks. Volatility measured on the quiet period will tell you the position is safe right up until it is not, so do not rely on a standard deviation model alone.

Track the whole basket as one exposure, not as a set of independent positions. In a stress event the correlation between high yielders goes to roughly one. Size the book on the assumption that every long position falls together.

Additional risks worth naming honestly. Capital controls can trap you in a position you cannot exit. NDF fixings can be suspended or disputed, meaning your hedge does not settle at the rate you expected. Forwards carry counterparty exposure to your bank. And in a real crisis the bid disappears entirely, so your loss is determined by where you can actually trade rather than where the screen says the market is.

Use expected shortfall or a scenario-based loss measure rather than plain value at risk. Value at risk built on quiet-period data systematically understates exactly the loss this strategy is exposed to.

Costs & implementation

Emerging market FX spreads are far wider than G10 spreads, and they widen further in stress, which is precisely when you may need to trade. Rolling forwards every month has a real cost. Where onshore markets are restricted you are trading non-deliverable forwards, which settle in dollars against a published fixing, and that fixing itself can be manipulated, delayed or suspended.

Budget for the option premium as a permanent drag of a meaningful share of gross carry. If the hedge does not visibly hurt returns in calm years, it is probably too far out of the money to help in bad ones.

Turnover is modest, but expect execution to be the difference between a backtest that works and a live book that does not.

Failure modes

• Chasing the highest yielder. The currency paying the most is usually paying the most for a reason, and the reason is often a devaluation that has not happened yet.
• Treating quiet periods as evidence of safety. Long stretches of calm are the normal state of this trade, not proof that the risk is gone.
• Under-hedging because the hedge feels wasteful. It always feels wasteful until the week it does not.
• Ignoring correlation. A book of eight high yielders is close to one position with eight names on it.
• Backtesting on survivors. Currencies that redenominated, pegged, collapsed or stopped trading must stay in the historical sample, or your results are fiction.
• Assuming you can exit at the last quoted price. In the events that matter, you cannot.

Our Notes & Suggestions

Treat this as a premium collection business, not as a directional view on any country. That framing changes the design decisions: you care more about position limits, hedging discipline and the exit rule than about picking the best currency.

A practical simplification for a smaller book is to run only the most liquid five or six emerging currencies, keep the hedge crude but constant, and accept a lower Sharpe ratio in exchange for a book that can actually be closed on a bad day. A carry strategy that cannot be exited is not a strategy, it is a bet on the calendar.

Finally, be honest in the reporting. Show the return with and without the worst month included. If the strategy only works when you remove the crash, then the crash is the strategy.