The Carry Trade Explained: Where the Real Money Lives in FX

The carry trade is the oldest, simplest, and most profitable strategy in foreign exchange. It is also the strategy that has destroyed more macro hedge funds than any other. Borrow a low-yielding currency, buy a high-yielding one, collect the rate differential. The mechanics fit on the back of a business card. The execution is where careers end. Every retail trader who learns FX eventually meets the carry trade and falls in love with the simplicity. Most lose money on it. The few who learn how to size, time, and exit it generate decades of compounding alpha. This guide is the institutional version of the carry trade. What it actually is, why it pays, why it breaks, the 2024 yen unwind, the modern G10 setup, and the framework professional desks use to extract risk-adjusted returns from one of the few genuinely persistent anomalies in FX.

By Ken Chigbo · Founder, KenMacro · 18+ years in markets, London trading floor and institutional FX

This guide is reviewed and refreshed periodically. The framework itself is timeless.

In one sentence: the carry trade is the systematic strategy of borrowing in a low-yielding currency to invest in a higher-yielding one, capturing the interest-rate differential as ongoing income, profitable for years at a time, then violently unwound in volatility spikes that wipe out months of gains in days, and the trick is not finding the trade but knowing when to be in it.

What the Carry Trade Actually Is

A carry trade is the systematic strategy of borrowing money in a currency where interest rates are low and using the proceeds to invest in a currency where interest rates are high. The trader pays the lower rate on the borrowed currency and earns the higher rate on the invested currency. The difference between the two is the carry. As long as the exchange rate between the two currencies stays roughly stable, the carry compounds month after month, year after year, with virtually no effort and no skill required beyond the initial setup.

The simplest possible example: borrow ten thousand Japanese yen at 0.5 percent, convert it to Australian dollars, and invest at 4.5 percent. The carry is roughly 4 percent per year, paid in interest, with the principal still subject to the exchange-rate movement of AUD/JPY. If AUD/JPY goes nowhere across the year, the trader collects 4 percent on the position. If AUD/JPY rises, the trader earns the carry plus capital gains. If AUD/JPY falls, the trader collects the carry but loses on the principal.

That structure has been the core of professional FX investing for fifty years. Carry-trade returns have outperformed broad equity indices on a Sharpe-ratio basis across many decades when measured across the full G10 currency basket. The Bank for International Settlements publishes regular research on the topic, including longer-term carry returns and cross-currency basis dynamics, at bis.org. The IMF has documented similar findings on emerging market carry. The strategy is not theoretical, it is empirical, and it has been a structural part of how global capital flows work since the 1970s.

For traders, the practical definition is sharper. If you can hold a long position in a high-yielding currency against a low-yielding one for months at a time and the rollover (overnight financing) credits your account night after night, you are running a carry trade. If you cannot hold for months because volatility forces you out, you are not really running carry, you are running directional FX with a tilt. The two strategies look identical on the screen and behave very differently in a drawdown.

The Mechanics: How a Carry Trade Makes Money

A carry trade has three sources of return and one source of risk. Understanding all four is the difference between an institutional approach and the retail-trader approach that loses money.

First source: the interest-rate differential, paid as overnight rollover. When you hold an FX position past the daily 17:00 New York rollover, your broker either credits or debits your account based on the difference between the two currencies' overnight interest rates. Long AUD/JPY when AUD pays 4.5 percent and JPY pays 0.5 percent earns roughly 4 percent annualised, paid in tiny daily increments. Long EUR/CHF when EUR pays 2.5 percent and CHF pays 0.75 percent earns roughly 1.75 percent annualised. The exact carry varies by broker spreads, but the rate-differential mechanism is universal.

Second source: capital gains on the exchange rate. Carry trades typically work in regimes where the high-yielding currency is also strengthening (because capital is flowing toward higher yields). So traders frequently earn both the carry AND the appreciation on the principal. The 2003 to 2007 cycle saw AUD/JPY rise from 60 to 105, a 75 percent capital gain on top of the carry, and that combined return is what made carry trades dominant in pre-crisis macro hedge fund portfolios.

Third source: the convexity premium during calm regimes. When volatility is low and carry is paying, there is a self-reinforcing dynamic where positions accumulate, broker financing costs fall, and the carry-on regime extends itself. This third-order effect is why carry tends to overshoot fundamentals. By the time the regime breaks, positioning is far more crowded than the fundamentals justify.

The risk: tail-risk drawdowns from the unwind. When carry breaks, it does so violently. Volatility spikes (typically driven by a global risk-off event, an unexpected central bank move, or a major fundamental shock to the high-yielder), positions get unwound across the global trading day in cascading liquidations, and high-yielding currencies can lose 5 to 15 percent in 48 hours. The carry trader who collected 4 percent over twelve months can give back the entire year's gain in two days. The risk profile is therefore positively skewed in normal markets (small steady gains) and negatively skewed in crisis markets (occasional huge losses). The expectation across many cycles is positive but the path is not.

Why Carry Works (and Why It Looks Like Free Money)

If carry trades reliably make money, why does the differential exist at all? Why don't markets simply close it through arbitrage until the carry is zero? Three reasons.

First, the carry trade is not actually free money in expectation. According to the Uncovered Interest Parity (UIP) hypothesis, the higher-yielding currency should depreciate over time by exactly the rate differential, perfectly offsetting the carry. If UIP held, the carry trade would have zero expected return. UIP fails empirically (this is one of the longest-running empirical anomalies in finance), but the failure is not permanent and it is not equally distributed. Carry compensates traders for taking on episodic crash risk, and the long-run expected return is positive but lumpy.

Second, capital is not perfectly mobile. Pension funds, central banks, and corporate treasuries cannot freely shift between currencies for tax, regulatory, and risk-management reasons. The pool of capital that can actively arbitrage the rate differential is large but limited, and that limit is what keeps the carry from being competed away.

Third, the risk of the carry trade is real and asymmetric. The carry pays small steady amounts most of the time and loses large amounts in occasional crashes. Insurance companies, pension funds, and other capital providers demand a premium to bear that risk profile. The carry differential is largely that premium. It is not an arbitrage; it is a risk-bearing return.

For practical traders, the takeaway is that carry trades work but are not free. The expected return across the cycle is positive but accompanied by tail-risk drawdowns. Sizing for the tail is the difference between earning the long-run premium and getting wiped out in a single unwind. Most retail traders learn this the hard way.

The Classic G10 Carry Pairs

The G10 currencies, the ten most liquid major currencies, dominate carry-trade activity because spreads are tight, liquidity is deep, and central bank policy is transparent. The classic carry pairs cluster around two funding currencies and two or three target currencies.

The most-traded carry pairs are AUD/JPY, NZD/JPY, AUD/CHF, NZD/CHF, and EUR/JPY. Each combines a structural funding currency with a structural target currency. AUD/JPY in particular has been the canonical "risk-on barometer" of FX for two decades, partly because it embodies the carry-on logic so cleanly: high-yielder against low-yielder, commodity-linked against safe-haven.

Beyond those classic pairs, traders rotate which currency plays which role based on the central bank cycle. When the Federal Reserve is hiking aggressively (as in 2022 to 2023), USD becomes a target currency rather than a neutral one, and pairs like USD/JPY become primary carry expressions. When the Bank of Japan finally exits zero rates (the 2024 cycle), JPY's funding role weakens. The roles are dynamic. The framework is permanent.

Emerging Market Carry: Higher Yield, Higher Pain

Beyond the G10, emerging market currencies offer dramatically higher carry but with proportionally larger tail risk. The Mexican peso, Brazilian real, Turkish lira, South African rand, and Indonesian rupiah have at various points offered carry of 8 to 25 percent per year against G10 funding currencies. That is more than three times the typical G10 carry.

The catch is that emerging market currencies are subject to political risk, central bank intervention, capital controls, and sudden devaluations that G10 currencies do not face. The 2018 Turkish lira crisis saw TRY lose 40 percent in three months. The 2015 Brazilian real meltdown took BRL down 30 percent in six months. The 2022 to 2024 Argentine peso effectively collapsed. Each of those events wiped out years of accumulated carry in weeks.

The professional rule on EM carry is to size positions much smaller than G10 carry, hedge with options where possible, and rotate out when local political or central bank stress builds. Most retail traders fail at all three. They size EM carry like G10 carry, do not hedge, and sit through the unwind. The result is consistently the same.

For most traders, the right answer is to stay in G10 carry until you have the institutional infrastructure (real-time political risk monitoring, options market access, portfolio-level VaR sizing) to handle EM. The G10 alpha is real and persistent without the EM tail risk.

The Hidden Cost: Vol-of-Vol and the Fat-Tail Problem

Realised carry-trade volatility looks low when measured day to day in calm markets. Annualised volatility for AUD/JPY in a normal regime might be 8 to 10 percent. That looks attractive against the 4 percent carry: a Sharpe ratio of 0.5 or higher, comparable to passive equity.

The catch is that the volatility itself is not stable. The volatility of volatility, often called vol-of-vol or VOV, is enormous in carry trades. Realised vol can sit at 8 percent for two years and then spike to 30 percent in a single week during an unwind. The standard volatility measure underweights this dramatically. Anyone sizing carry positions on rolling 30-day realised volatility will systematically overweight the trade exactly when the regime is most fragile.

The cleanest way to think about this is in terms of the return distribution. Carry returns are positively skewed in calm regimes (small consistent gains) and negatively skewed in crisis regimes (occasional very large losses). When you average across cycles, the mean is positive but the path is asymmetric. Standard deviation is a poor risk measure for this distribution. Value-at-Risk at the 99 percent confidence level, conditional on a vol regime change, is the institutional measure.

For practical traders, this means three things. First, do not size carry positions based on recent realised volatility. Second, always assume a 5 to 10 percent overnight gap risk in a worst-case unwind. Third, use options or stop-out levels to cap downside, even at the cost of giving up some carry. The professional desks use VOV as the regime indicator: when implied vol of FX options spikes from low to high, position sizes are cut aggressively regardless of the spot move.

When the Carry Trade Breaks: The Unwind Mechanics

Carry trades unwind in a predictable cascade once the trigger fires. The mechanics are worth memorising because they repeat across cycles with mathematical precision.

Each unwind looks slightly different in detail but the cascade structure is universal. The 1998 Long Term Capital Management unwind, the 2008 Lehman crisis, the 2015 Swiss franc removal of the EUR/CHF floor, the 2020 March pandemic crash, and the 2024 yen carry unwind all followed this five-step pattern with different triggers and different magnitudes. The pros do not try to predict each trigger. They watch the second-step indicator (FX implied volatility) and exit when it spikes regardless of the news cycle.

The 2024 Yen Carry Unwind: A Modern Case Study

The August 2024 yen carry unwind is the most recent textbook example and is worth studying in detail because every element of the cascade played out in a compressed timeline that highlighted exactly how the mechanics work.

The trigger fired on 31 July 2024 when the Bank of Japan, ending two decades of de facto zero rates, hiked the policy rate by 15 basis points to 0.25 percent and signalled further hikes ahead. The same week, US Non-Farm Payrolls came in 64,000 below consensus and triggered the so-called Sahm Rule recession indicator. Two simultaneous shocks: the funding currency suddenly cost more, and the target environment looked riskier.

USD/JPY fell from 153 on 1 August to 142 on 5 August, a 7.2 percent move in three trading days. AUD/JPY fell from 100 to 90, a 10 percent move. The Nikkei dropped 12 percent on 5 August alone, the largest single-day fall since 1987. The VIX spiked from 16 to 65 intraday, the second-highest reading in history. Japanese government bond yields whipsawed. Global equity markets fell 3 to 6 percent in sympathy as risk parity funds rebalanced.

By 8 August, the unwind had exhausted. USD/JPY stabilised around 145, AUD/JPY around 93, the VIX returned to the low 20s. The total peak-to-trough drawdown for a typical AUD/JPY carry position was 12 to 15 percent. For a trader who had entered the carry in March 2024 with AUD/JPY at 95, six months of carry returns (roughly 2 percent in interest) were wiped out in three days, plus a 7 percent capital loss. The Bank of Japan published its full policy statement and accompanying analysis at boj.or.jp the same day. The IMF subsequently issued a Working Paper analysing the cross-border financial stability dimensions of the unwind, available through their research portal at imf.org. Both are essential reading for any serious carry trader, because the institutional analysis from those primary sources captures the cascade dynamics in ways the financial press systematically misses.

The lessons from August 2024 are now standard institutional teaching. First, the trigger was a known potential risk: the BoJ had been signalling hikes for months. Second, the cascade ran exactly through the five-step structure. Third, the recovery was fast (within 6 weeks AUD/JPY had retraced half the unwind), but the traders who held through suffered the full drawdown while those who exited at step two re-entered at the better levels. Fourth, vol-of-vol was the early warning indicator: implied volatility on USD/JPY options had risen meaningfully in mid-July before spot reacted. The signal was visible to anyone watching options markets. Most spot-only traders missed it.

How Real-Yield Differentials Drive Carry

The textbook carry trade is described in nominal terms: borrow at 0.5 percent, invest at 4.5 percent, capture 4 percent. The institutional version is in real terms: borrow at minus 1 percent real, invest at plus 1 percent real, capture 2 percent real carry. The two framings give very different signals about which trades to enter, when to enter them, and when to exit.

Capital allocators care about real returns, not nominal returns. International investors comparing investment options in dollars, yen, and euros do not just compare nominal yields; they compare expected real yields after adjusting for each currency's expected inflation. A nominal-yield differential of 4 percent that is entirely driven by higher Australian inflation gives no real carry advantage. A real-yield differential of 2 percent gives genuine real-purchasing-power compensation for currency risk.

Real yields are derived from inflation-protected sovereign debt: TIPS in the US, JGBi in Japan, ILBs in Australia. The full mechanics of the real-yield curve and how it drives every cross-asset including FX is unpacked in the real yields explained guide. For carry traders, the simplest rule is: track real-yield differentials, not nominal-yield differentials, when sizing carry positions. When a nominal differential is widening but the real differential is flat or narrowing, the carry trade is weaker than it looks.

The 2026 setup is informative. Nominal AUD-JPY differential is around 4 percent. But Australian inflation is also higher than Japanese inflation, so the real-yield differential is closer to 1.5 to 2 percent. The carry trade is real but smaller than the nominal print suggests. A trader sizing on nominal alone is over-sizing relative to the actual risk-adjusted compensation.

Carry vs Trend: The Two FX Trading Religions

FX traders sort into two camps: carry traders and trend traders. Each has its religion, its champions, and its preferred regimes. Knowing which camp you are in (or whether to alternate) is essential to long-term success.

Carry traders make money in low-volatility, range-bound markets. They collect the rate differential, sit through small moves, and accept episodic large drawdowns as the cost of doing business. The strategy works because most of the time, FX exchange rates do not move enough to wipe out the carry. The classic carry trader runs a long-AUD/JPY position for two years, collects 8 percent per year, and gives back 6 percent in a single unwind. Net return: 10 percent over two years, with a Sharpe of 0.6 to 1.0 across cycles.

Trend traders make money in trending, directional markets. They do not care about interest-rate differentials. They care about momentum. The classic trend trader follows medium-term moving averages, scales into a position as it confirms the trend, and rides it for 3 to 12 months until the trend breaks. The strategy works because FX has structural trends that persist longer than random walks would predict. The classic trend trader catches the 2014-2015 USD bull market, the 2022 EUR collapse, the 2024 yen unwind, and accepts long flat periods between trends.

The two strategies are statistically uncorrelated, which means they combine beautifully. The institutional approach is to run both side-by-side, sized inversely to their recent performance, with a rebalance every six months. Hedge funds like Bridgewater and AQR have done this systematically for decades with strong risk-adjusted returns. Retail traders who pick one and stick with it through wrong-regime periods underperform both strategies combined.

The Three Carry-Trade Regimes

Across cycles, the carry environment lives in one of three regimes. Identifying which regime you are in is more important than picking the right pair, because regime determines whether the trade pays at all.

Carry-on regime. Volatility is low, central bank policies are stable, and risk assets are rising. Real-yield differentials are wide. The classic 2003 to 2007 cycle, the 2017 cycle, and parts of 2019 to 2021 were all carry-on regimes. Sharpe ratios on G10 carry baskets ran above 1.0 in those windows. Position sizes can be larger, hedging costs lower, and the strategy compounds steadily.

Carry-off regime. Volatility is spiking, central banks are surprising, and risk assets are correcting. Funding currencies rally hard against target currencies. The 2008 crisis, the 2020 pandemic crash, the August 2024 yen unwind, and parts of 2022 were all carry-off regimes. Carry trades lose 5 to 15 percent in days. Position sizes should be zero or negative (i.e. short the carry trade by long-funding-versus-target).

Low-carry range regime. Volatility is moderate, real-yield differentials are narrow, and conditions are mixed. Profitable carry trades are smaller and harder to identify. Most institutional desks stay flat or reduce exposure significantly. Cycles like late 2010 to 2012 and 2018 fit this profile. Selective pair-by-pair work can extract returns, but the systematic carry basket struggles.

The discipline is to read which regime you are in and adjust position size accordingly. The biggest mistake retail carry traders make is treating all regimes as carry-on. The biggest mistake institutional desks make is treating regime transitions as gradual when they are actually abrupt.

Cross-Asset Impact of Carry Regimes

Asset by Asset Carry Map

Scenario Map: When Carry Works and When It Breaks

The KenMacro Carry Framework

The framework is four steps. Run it weekly during carry regimes, daily during transitions.

First step: classify the regime. Is implied volatility low and falling, range-bound, or rising? Use VIX and FX implied volatility (DXY 1-month implied) as the primary indicators. Below 14 VIX with falling implied vol is carry-on. Above 18 VIX with rising vol is carry-off. In between is range. The classification dictates position size more than any other variable.

Second step: rank the pairs. For each candidate carry pair, calculate the real-yield differential (nominal yield minus expected inflation), not just the nominal differential. Rank pairs by real-yield differential. Trade the top three to five pairs only. Skip pairs where the real differential is below 1 percent regardless of how attractive the nominal looks.

Third step: size for the tail. Position size should be calibrated against the worst historical drawdown for the pair, not against recent realised volatility. AUD/JPY has had 15 percent overnight drawdowns in cascades. Size positions so that a 15 percent overnight gap on the pair would not cost more than 2 percent of total portfolio NAV. That mathematical exercise is what separates institutional sizing from retail sizing.

Fourth step: monitor the exit. Watch FX implied vol daily. The signal to exit is a 30 percent or larger increase in 1-month implied volatility within 24 hours. Do not wait for spot to reverse. By the time spot reverses, you are in the cascade. The vol signal leads the spot signal by 24 to 72 hours in most unwinds. Use it.

Trader Playbook

Common Carry-Trade Mistakes

First mistake: sizing on recent realised volatility. Carry trades look low-vol most of the time. Sizing on rolling 30-day vol systematically overweights the trade exactly when the regime is most fragile. The fix is to size on conditional VaR or against the worst historical drawdown for the pair, not against recent realised.

Second mistake: ignoring vol-of-vol. Realised volatility lags implied volatility in carry unwinds. By the time realised vol rises, the cascade is well underway. Track FX implied volatility (USD/JPY 1-month, AUD/USD 1-month) as the leading indicator. A 30 percent rise in implied vol within 24 hours is the early warning signal to exit.

Third mistake: confusing nominal differentials with real differentials. Nominal-yield differentials can be wide while real-yield differentials are narrow if the high-yielder also has higher inflation. The carry trade is a real-yield trade. Trade the real differential, not the nominal print. The full real-yield framework is in the real yields explained guide.

Fourth mistake: holding through unwinds. Retail traders watch the cascade and freeze. Professional desks exit at step two (the vol spike). The difference is the entire profit and loss across cycles. Practice exiting before you need to. Set rules. Pre-commit to the exit criteria before the regime shifts.

Fifth mistake: oversizing emerging markets. EM carry looks better than G10 carry by 5 to 15 percent annualised. The 2018 Turkish lira lost 40 percent in three months. The 2022 Argentine peso effectively collapsed. Size EM carry at one-third to one-half of G10 carry. The math on tail risk demands it.

Sixth mistake: trading carry without macro context. The Fed cycle, the BoJ cycle, the geopolitical environment, all condition the carry regime. Ignoring the macro and trading carry pairs in isolation produces inconsistent results. The institutional approach is to embed carry into a broader macro framework. The how to trade FOMC guide unpacks how Fed cycles drive carry conditions across both target and funding currencies.

What Would Invalidate the Carry Framework

Final Takeaway: Carry Pays, But Only If You Survive the Unwinds

The carry trade is one of the few persistent positive-expected-return strategies in macro. Across the long run, G10 carry baskets have outperformed broad equity indices on a Sharpe-ratio basis. The strategy is real, the alpha is real, and the academic and empirical literature confirms both.

What it is not is easy. Carry pays steadily for 18 to 36 months at a time, then violently gives back 8 to 15 percent in 3 to 10 days during unwinds. The traders who win across cycles are the ones who recognise the regime, size for the tail, exit at the vol-spike step rather than the spot-reversal step, and re-enter once volatility stabilises. The retail approach of treating carry as a buy-and-hold trade systematically loses money because the unwinds wipe out the carry returns.

For traders who want to extract the carry premium without the binary tail risk, the discipline is regime classification (carry-on, range, carry-off), real-yield-based pair selection, tail-sized positioning, and pre-committed exit criteria based on implied volatility. The framework is teachable, the rules are simple, the execution is the hard part.

The carry trade is where the real money lives in FX. It is also where careers go to die. The difference between the two is institutional discipline applied to a strategy that looks deceptively simple from the outside.

Frequently Asked Questions: The Carry Trade

Related institutional reading from the desk

The desk's deepest pieces on the macro framework, broker selection, and prop firm economics for serious traders.

• Best Forex Broker for Day Trading 2026, KenMacro Top 3 Verdict
• Best Forex Broker for Macro Trading 2026
• Best Forex Broker for Emerging Markets 2026, Star Trader vs Exness
• Best Prop Firm for Macro Traders 2026
• How Much Do Prop Firm Traders Actually Make in 2026
• How to Trade Gold (XAUUSD), the Macro Trader's Institutional Guide
• How to Trade NFP, the Macro Trader's Guide
• Vantage vs Blueberry vs Star Trader 2026, Three-Way Verdict

For the live framework that runs every London open, the desk's macro intelligence layer at KenMacro publishes daily. Free Discord access and full archive at kenmacro.com.