Matrixx0070 FX Carry Trade Analysis
When to use
Use this to evaluate the carry available in a currency pair or a basket: the interest-rate differential, the forward points implied by covered interest parity, the expected carry return net of the forward drag, the carry-to-volatility ratio, and the crash/skew risk that makes carry a "picking up pennies" strategy. Works for single pairs and for ranked G10/EM carry baskets.
Not for: rates-curve views on a single currency (lseg-swap-curve-strategy), broad macro regime monitoring (lseg-macro-rates-monitor), or option-vol surface analysis (lseg-option-vol-analysis) — though carry risk is fundamentally an options/skew story, so cross-reference it.
This is educational analysis, not investment advice.
Method
- Get the rate differential. Carry ≈ (funding-currency rate − investment-currency rate is negative carry; invest high-yield, fund low-yield for positive). Use comparable-tenor money-market or OIS rates, same day-count.
- Check covered interest parity (CIP). Forward points should equal the differential:
F = S × (1+r_quote·τ)/(1+r_base·τ). Compute the CIP-implied forward and compare to the market forward — the gap is the cross-currency basis (a real, persistent dislocation, not free arbitrage). - Compute the carry return correctly. Unhedged expected carry over horizon τ ≈ rate differential − expected spot depreciation of the high-yielder. The forward already prices the differential (that's CIP), so a pure forward position earns carry only if spot does not move to the forward — i.e. carry is a bet against uncovered interest parity (UIP).
- Normalize by volatility. Compute carry-to-vol = (annualized carry) / (annualized FX vol). This is the Sharpe-like screen; 3% carry at 6% vol (0.5) beats 5% carry at 15% vol (0.33).
- Quantify crash risk. Carry currencies have negative skew — they "go up by the stairs, down by the elevator." Read the FX risk-reversal (25-delta) as the market's skew price; a deeply negative risk-reversal on the high-yielder signals expensive crash insurance and crowded carry.
- Rank / size (basket). Rank pairs by carry-to-vol, go long the top / short the bottom, size inversely to vol so each contributes equal risk. Note correlation — carry trades are correlated (they all sell off together in risk-off).
- State the drawdown risk explicitly. Carry's return distribution is fat-left-tailed; report the historical max drawdown and the fact that positive carry accrues slowly and reverses violently.
Example
Long MXN / short JPY, 3-month. MXN 3m rate ≈ 10.8%, JPY ≈ 0.4% → differential ≈ +10.4% annualized carry. Forward points confirm via CIP (small negative cross-currency basis). 3m implied vol ≈ 9.5% → carry-to-vol ≈ 1.09 — attractive on the screen. BUT the 25-delta risk-reversal favors MXN puts by 2.8 vol points (market pricing MXN crash risk), and the pair's historical max drawdown in risk-off episodes exceeds −15% in weeks. So the 1.09 ratio understates tail risk. Read: carry is real and well-compensated on average, but this is a short-vol / short-skew position — size for the crash, not the average. Illustrative — NOT advice.
Pitfalls
- Double-counting carry via the forward. Buying the high-yielder forward does not add carry on top of spot's differential — the forward is the differential (CIP). Carry P&L comes from spot failing to converge to the forward (UIP violation).
- Ignoring the cross-currency basis. For funded/hedged trades the basis is a real cost or pickup, especially in USD funding stress — don't assume clean CIP.
- Screening on raw carry, not carry-to-vol. High nominal carry usually just means high vol; normalize.
- Treating carry as normally distributed. It has strong negative skew and fat left tails — VaR from a normal assumption badly understates crash risk.
- Crowding blindness. When everyone is in the same carry trade, positioning unwinds amplify the crash — watch risk-reversals and positioning data.
- EM settlement/liquidity gaps — NDFs, capital controls, and holiday calendars change the realized return.
Output format
Pair / basket: <long high-yield / short low-yield> | Horizon: <..>
Rate differential: <r_high − r_low> = <..>% annualized (tenor/day-count noted)
CIP check: implied fwd <..> vs market fwd <..> → cross-ccy basis <..>bp
Carry return: <..>% (note: earned only if spot ≠ forward — a UIP bet)
Volatility: <..>% | Carry-to-vol: <..>
Skew/crash: 25Δ risk-reversal <..> vol pts | historical max drawdown <..>%
Sizing: inverse-vol, correlation caveat <..>
Read: carry <attractive/thin> per unit risk; tail-risk warning: <..>
NOT investment advice — educational analysis only.
Reference
Covered interest parity (CIP)
F = S × (1 + r_quote·τ) / (1 + r_base·τ) (quote = price currency, e.g. the "quote" in EURUSD is USD). Forward points = F − S. CIP is enforced by arbitrage for freely-tradable currencies, so the forward mechanically embeds the rate differential. Persistent deviations = the cross-currency basis, driven by USD funding demand, bank balance-sheet costs, and regulation — real, not arbitrageable for most participants.
Uncovered interest parity (UIP) and why carry works
UIP claims the high-yielder should depreciate by exactly the differential, making carry zero in expectation. Empirically UIP fails at short/medium horizons — high-yielders depreciate less than UIP predicts (often appreciate), so carry earns a positive average return. That excess return is compensation for crash risk, not a free lunch (the "forward premium puzzle").
Carry-to-vol (the screen)
Carry-to-vol = annualized rate differential / annualized FX volatility. A Sharpe-like ranking. Use realized or implied vol consistently. In a basket, size positions inversely to vol so each pair contributes equal risk budget; then account for the high positive correlation among carry pairs (they are one macro factor).
Skew, risk-reversals, and the crash
FX risk-reversal = IV(call) − IV(put) at a delta (e.g. 25Δ). A negative risk-reversal on the high-yielder means puts (crash protection) are bid — the market prices asymmetric downside. Carry returns have negative skew and excess kurtosis; the strategy is economically short a put on the high-yielder. Monitor risk-reversals as a real-time crash-risk and crowding gauge.
Hedged vs unhedged
- Unhedged: full spot exposure; earns the differential if UIP fails, bears the crash.
- Fully hedged (rolled forwards): locks the forward, so it earns only the cross-currency basis — carry is hedged away by construction. "Carry" as a strategy is inherently an unhedged/partially-hedged spot bet.
Basket construction and factor risk
Long top-carry / short bottom-carry across G10 or EM diversifies idiosyncratic FX moves but concentrates the carry factor — highly correlated to global risk appetite (VIX, credit spreads). Treat aggregate carry exposure as a single risk-on beta and stress it against a risk-off shock, not as independent pair bets.