AI Trading Strategies / Carry Trade (FX)

Carry Trade (FX): AI-Powered Analysis Without Excel Hell

The FX carry trade is brilliantly simple: borrow low, lend high, pocket the spread. USD at 5.50%, JPY at -0.10%—that's 560 basis points of free money. Until currency volatility wipes out six months of gains in three days. Here's how AI turns carry trade analysis from spreadsheet agony into instant intelligence.

Andrew Grosser

Andrew Grosser

February 24, 2026 • 14 min read

November 2023: Fed funds at 5.50%, Bank of Japan at -0.10%. You're running a $2M carry portfolio across five pairs: USD/JPY (560bp differential), AUD/JPY (445bp), NZD/JPY (535bp), BRL/USD (875bp), TRY/USD (2,400bp—yes, really). Daily rollover on USD/JPY alone: $306 per $1M notional. Annualized carry: 11.2% if exchange rates stay flat. But they never stay flat. December 2023: Risk-off event, yen rallies 3.2% in 48 hours, erasing $64,000 from your USD/JPY position—21 days of carry income gone. Your Excel tracking needs: real-time spot rates for all pairs, forward points to calculate implied yield, historical correlation matrices (do all five pairs crash together?), volatility-adjusted position sizing, and daily P&L reconciliation against broker statements showing tom-next swap rates that never quite match your calculations.

Excel makes carry trade management brutal. You're pulling spot rates from GOOGLEFINANCE() that go stale on weekends, calculating annualized carry using ((1+r_high)/(1+r_low)-1)×(365/days), building correlation matrices with MMULT() that crash when you add a sixth pair, tracking rollover P&L that varies by broker and updates at 5pm EST daily, and stress-testing scenarios where VIX spikes cause funding currencies to rally 5% overnight. Change TRY rates from 24% to 25% after Erdogan fires another central banker, and watch #REF! errors cascade through 6 worksheets. Sourcetable eliminates this nightmare. Upload your positions with spot rates and interest rate differentials, ask "What's my total daily carry?" Get instant aggregated income. Request "Show correlation between my pairs during high VIX periods" and see exactly how concentrated your tail risk is. sign up free.

What Makes Carry Trades So Hard to Analyze

The carry trade concept is deceptively simple: borrow in a low-rate currency (funding currency), convert to a high-rate currency (target currency), invest at the higher rate, and capture the interest differential. If USD yields 5.50% and JPY yields -0.10%, you earn 5.60% annually on the spread—assuming exchange rates stay constant.

But exchange rates never stay constant. Currency volatility can dwarf interest income. A 2% adverse move in USD/JPY wipes out 4 months of carry. A 5% move (common during risk-off events) destroys a year of gains. This creates the fundamental carry trade challenge: steady small gains punctuated by catastrophic losses—the classic "picking up pennies in front of a steamroller" problem.

Sophisticated carry analysis requires tracking multiple dimensions:

  • Interest rate differentials across dozens of currency pairs
  • Forward rates to see what the market expects (covered interest parity)
  • Implied yield from forward points vs. spot rates
  • Daily rollover rates (tom-next swaps) that brokers actually charge
  • Historical volatility to size positions appropriately
  • Correlation matrices to assess diversification
  • Central bank policy calendars for upcoming rate decisions
  • Risk-off indicators (VIX, credit spreads) that predict funding currency strength

In Excel, this becomes 8 separate workbooks with hundreds of cross-referenced formulas that break when markets are closed or data providers have outages. Managing five carry pairs requires 40+ manual calculations daily just to track P&L accurately.

How Sourcetable Turns Carry Analysis Into Natural Language

Sourcetable doesn't eliminate the complexity of carry trades—it eliminates the manual labor of analyzing them. Upload your positions, spot rates, and interest rate data, then ask questions the way you'd ask a trading desk analyst.

Instant Interest Differential Calculations Across All Pairs

In Excel, calculating carry for USD/JPY requires: spot rate (149.00), USD rate (5.50%), JPY rate (-0.10%), position size ($1M), then: daily_carry = position × (rate_diff / 365) / spot = $1M × 0.056 / 365 / 149 = $306. Now multiply by five currency pairs with different conventions (some quote USD first, some don't), and you're building lookup tables with VLOOKUP chains.

In Sourcetable, upload a table with columns: Pair, Position Size, Spot Rate, High Rate, Low Rate. Ask: "Calculate daily carry for all positions." The AI instantly returns a new column showing $306 for USD/JPY, $187 for AUD/JPY, $221 for NZD/JPY—total daily carry: $714. Ask: "Which pair has the highest yield?" It highlights TRY/USD at $1,314 daily—but flags the 18% annualized volatility risk. No formulas written.

Automatic Forward Rate Implied Yield Analysis

Covered interest parity says forward rates should reflect interest differentials. If they don't, arbitrage opportunities exist—or the market expects currency moves. Forward points for 3-month USD/JPY: +210 pips. Spot: 149.00. Forward: 151.10. Implied appreciation: 1.41% quarterly = 5.64% annualized. This matches the 5.60% rate differential—no free lunch.

Calculating this manually requires: forward_yield = ((forward/spot - 1) × (365/days)). Do this for five pairs with different tenors (1M, 3M, 6M), and you're writing nested formulas tracking contract dates and day count conventions.

Ask Sourcetable: "What's the implied yield from forward rates?" It calculates across all your pairs and tenors, returning: USD/JPY 3M forward implies 5.61% (matches carry—fair pricing), but AUD/JPY implies only 3.8% (market expects yen strength, reducing carry attractiveness). This kind of arbitrage scanning takes Excel users hours. Sourcetable does it in seconds.

Real-Time Correlation and Risk Concentration Analysis

The 2008 crisis taught carry traders a brutal lesson: during risk-off events, all funding currencies strengthen simultaneously. JPY, CHF, and USD (when it's the funding currency) rally together, crushing diversified carry portfolios. Effective risk management requires understanding correlation during stress periods, not normal times.

In Excel, this means: download 60 days of returns for all pairs, calculate correlation matrix using CORREL() across every pair combination, filter for high-VIX days (VIX > 25), recalculate correlations for just those days. That's 10 pairwise correlations for 5 currency pairs, requiring 10 separate correlation formulas.

Ask Sourcetable: "Show correlation matrix when VIX exceeds 30." It filters historical data to stress periods, calculates all pairwise correlations, and returns a heatmap showing AUD/JPY and NZD/JPY correlate at 0.89 during panics (you're not diversified), but BRL/USD only correlates at 0.42 (genuine diversification). The AI explains: "Your AUD and NZD positions move together during selloffs—consider reducing one." Actionable risk management delivered conversationally.

Scenario Analysis and Stress Testing

Carry traders need to know: "If yen rallies 5% overnight (like March 2020), how much do I lose?" Excel requires building scenario tables with multiple exchange rate assumptions, calculating P&L for each, and manually comparing outcomes.

Ask Sourcetable: "What's my P&L if JPY strengthens 5% across all pairs?" It calculates instantly: USD/JPY loses $50,000, AUD/JPY loses $35,000, NZD/JPY loses $28,000—total drawdown $113,000, wiping out 158 days of carry income. Follow-up: "How much carry would I lose if I reduced exposure by 30%?""$214 daily vs. current $306, but max loss drops to $79,000—better risk-reward." This kind of dynamic scenario modeling would require rebuilding Excel models multiple times.

Managing Multiple Carry Positions as a Portfolio

Individual carry trades are straightforward. Managing 5-10 pairs simultaneously—tracking daily P&L, rebalancing as correlations shift, and adjusting to central bank policy changes—becomes operationally complex. Sourcetable centralizes everything.

Aggregated Position Tracking and Daily Reconciliation

Your broker statement shows: USD/JPY earned $318 yesterday (not the $306 you calculated), AUD/JPY earned $192 (vs. $187 expected). Differences come from exact tom-next swap rates, slight spot rate timing differences, and broker spreads. Reconciling actual vs. expected across five pairs daily is tedious but essential.

Upload broker statements alongside your expected P&L calculations. Ask: "Show me the variance between expected and actual carry." Sourcetable highlights: USD/JPY +$12 (broker swap rate was slightly favorable), TRY/USD -$47 (higher-than-expected funding cost). You immediately see where broker costs are eating returns and can negotiate better swap rates or switch brokers.

Dynamic Rebalancing Based on Changing Conditions

Central banks change policy constantly. The RBA hikes 25bp, shrinking AUD/JPY carry from 445bp to 420bp. Do you reduce exposure or maintain? The decision depends on how much you're earning elsewhere and whether volatility increased (negating the lower carry).

Ask Sourcetable: "Rerank my positions by risk-adjusted carry after today's rate change." It recalculates: NZD/JPY now offers better Sharpe than AUD/JPY (535bp carry with only 12% vol vs. 420bp with 11% vol). The AI suggests: "Consider shifting $200K from AUD to NZD to improve risk-adjusted returns." Portfolio optimization delivered as actionable advice.

When Carry Trades Work (and When They Explode)

Understanding when to deploy carry trades—and when to get out—is the difference between steady profits and catastrophic losses.

Best Conditions for Carry Trades

  • Low volatility environments: VIX below 15, stable FX markets, no major geopolitical shocks. Carry income accrues steadily without violent currency swings.

  • Diverging central bank policies: When one central bank is hiking (Fed) while another is cutting or holding (BOJ), rate differentials widen, increasing carry potential.

  • Positive risk sentiment: S&P near all-time highs, credit spreads tight, investors seeking yield. Capital flows from low-rate to high-rate currencies, supporting your positions.

  • Stable or favorable exchange rate trends: If the target currency is also appreciating (carry + FX gains), returns compound beautifully.

When to Avoid or Exit Carry Trades

  • Risk-off events: VIX spikes above 25, credit spreads widening, geopolitical shocks. Funding currencies (JPY, CHF, USD) rally violently as investors unwind risk.

  • Central bank policy convergence: When rate differentials start narrowing (high-rate central bank cuts, low-rate central bank hikes), carry attractiveness disappears.

  • Extreme positioning: When everyone is long the same carry trade (AUD/JPY in 2007), even small selloffs trigger massive liquidations as overleveraged traders hit stop losses.

  • Funding currency intervention risk: BOJ or SNB intervention to weaken their currency can cause sudden moves that wipe out months of carry overnight.

Sourcetable helps identify regime shifts. Connect VIX data and ask: "When VIX exceeded 20 over the past year, how did my carry pairs perform?" It shows you: every VIX spike correlated with JPY strength and carry losses. This creates a clear exit rule: reduce carry exposure 50% when VIX crosses 20. Backtest this: "What would my returns be with a VIX-based exit rule?" See immediately whether risk management improves Sharpe ratios.

Key Takeaways

  • The FX carry trade captures interest rate differentials by borrowing low-rate currencies and investing in high-rate currencies. Steady income with tail risk from currency volatility.

  • Traditional Excel analysis requires tracking spot rates, forward rates, tom-next swaps, correlation matrices, and stress scenarios across multiple pairs—hundreds of formulas that break constantly.

  • Sourcetable turns carry analysis into plain English: "What's my daily carry?" → $714. "Show correlation during stress." → Heatmap showing 0.89 correlation between AUD/JPY and NZD/JPY during panics.

  • Best conditions: low VIX, diverging central bank policies, positive risk sentiment. Worst conditions: risk-off events, policy convergence, extreme positioning.

  • Risk management is paramount: carry trades deliver small steady gains until they don't. Use VIX thresholds, correlation analysis, and position sizing to survive the inevitable blow-ups.

Frequently Asked Questions

If your question is not covered here, you can contact our team.

Contact Us
What is a carry trade in forex?
A carry trade involves borrowing money in a currency with low interest rates (funding currency like JPY at -0.10%), converting to a currency with high interest rates (target currency like USD at 5.50%), and investing at the higher rate. You earn the interest rate differential (5.60% in this example) as long as exchange rates don't move against you. The strategy generates steady income but carries significant currency risk.
How do you calculate carry trade returns?
Daily carry = (Position Size × Interest Rate Differential / 365) ÷ Spot Rate. For example: $1M position in USD/JPY with 5.60% differential at 149.00 spot = ($1M × 0.056 / 365) / 149 = $306 daily or $111,690 annually. Total return also includes exchange rate changes—if USD appreciates 2%, you gain an additional $20K. If it depreciates 2%, you lose $20K despite earning carry.
What are the biggest risks in carry trading?
Currency volatility is the primary risk. A 5% adverse move in exchange rates can wipe out a year of carry income. Risk-off events cause funding currencies (JPY, CHF) to rally sharply as traders unwind positions simultaneously. Leverage amplifies losses—many retail traders use 10:1 or higher leverage, turning a 5% currency move into a 50% account loss. Central bank intervention and sudden policy changes also create tail risks.
Which currency pairs are best for carry trades?
High-rate currencies (AUD, NZD, BRL, TRY, MXN) funded by low-rate currencies (JPY, CHF, EUR historically) offer the largest differentials. USD/JPY is the most liquid and lowest spread. AUD/JPY and NZD/JPY are popular but highly correlated. Emerging market pairs like BRL/USD or TRY/USD offer higher carry but much higher volatility. Best pair depends on your risk tolerance and correlation with existing positions.
How do you manage carry trade risk?
Position sizing is critical—risk no more than 1-2% of capital per pair. Diversify across low-correlation pairs (don't stack multiple JPY-funded trades). Use stop losses at technical levels or volatility-based exits (e.g., reduce exposure 50% when VIX exceeds 20). Monitor correlation during stress periods—pairs that seem diversified in calm markets often correlate 0.8+ during selloffs. Never use maximum leverage—carry trades require surviving drawdowns to collect long-term income.
When should you exit carry trades?
Exit when: (1) Central banks converge policy (rate differentials narrow), (2) VIX spikes above 25-30 (risk-off environment), (3) funding currency breaks key technical resistance (JPY rally), (4) you've captured 70%+ of potential carry for the year (lock in gains), or (5) correlation between your positions exceeds 0.7 (concentrated risk). Don't hold through known risk events like central bank meetings unless you accept potential overnight losses.
What returns can you expect from carry trading?
Skilled carry traders target 8-15% annual returns with 10-15% volatility in balanced portfolios. Individual pairs may offer higher nominal carry (TRY/USD: 24%+) but come with 25-35% volatility. The strategy performs best in low-volatility environments—2004-2007 saw consistent double-digit returns. It performs worst during crisis periods—2008 saw drawdowns exceeding 50% for leveraged carry portfolios. Long-term Sharpe ratios of 0.6-0.8 are realistic for disciplined strategies.
Andrew Grosser

Andrew Grosser

Founder, CTO @ Sourcetable

Sourcetable is the AI-powered spreadsheet that helps traders, analysts, and finance teams hypothesize, evaluate, validate, and iterate on trading strategies without writing code.

Share this article

Sourcetable Logo
Ready to master FX carry trades with AI?

Analyze interest differentials in seconds, not spreadsheets. Track correlations, stress-test scenarios, and optimize positions—all in plain English.

Drop CSV