Carry Strategies: AQR's Framework for Harvesting Yield Across Asset Classes

"Carry" is the return from holding an asset assuming its price doesn't change. A Japanese yen bond yields 0.1%. An Australian dollar bond yields 4.3%. Borrow yen, buy Australian dollars, pocket the 4.2% spread. Simple concept, enormous industry: AQR's carry fund manages $8B, Two Sigma runs $15B+ in multi-asset carry. Across FX, fixed income, equities, and commodities, the carry premium has averaged 4-7% annually with low equity correlation—but with the tail risk of sudden, severe drawdowns when risk appetite collapses.

Executive Summary

What Is Carry?

• The return earned from holding an investment unchanged—the "income" component before price appreciation
• Carry = forward price yield - spot price yield (simplification: yield differential)
• Positive carry: You earn by holding. Negative carry: You pay to hold.
• Carry strategy: Systematically long high-carry assets, short low-carry assets (within each asset class)

Carry Across Asset Classes:

• FX Carry: Borrow low-rate currencies (JPY, CHF), invest in high-rate currencies (BRL, MXN, AUD). Classic "carry trade."
• Bond Carry: Ride the yield curve (buy 10Y, hedge duration risk, earn roll-down premium)
• Equity Carry: Long high-dividend/buyback yield stocks, short low-yield stocks
• Commodity Carry: Long backwardated commodities (spot > futures = positive roll), short contango commodities
• Volatility Carry (VRP): Sell implied vol (overpriced), earn premium vs. realized vol

Part 1: FX Carry — The Original Carry Trade

Uncovered Interest Rate Parity: The Theory That Fails

Economic theory says interest rate differentials should be offset by currency movements. If Australia offers 4% and Japan offers 0%, the Australian dollar should depreciate 4% against the yen per year—leaving investors indifferent.

Reality: This theory (Uncovered Interest Rate Parity, UIP) fails empirically. High-rate currencies don't depreciate—they often appreciate slightly. This UIP failure is the source of FX carry returns.

30-year FX carry performance (G10 currencies):

The Classic FX Carry Portfolio (2026 Example)

Long (high-rate) currencies:

• Mexican Peso (MXN): ~9.5% interest rate → 8% carry above USD
• Brazilian Real (BRL): ~10.5% rate → 9% carry
• South African Rand (ZAR): ~8.2% rate → 6.7% carry
• Norwegian Krone (NOK): ~4.0% rate → 2.5% carry

Short (low-rate) currencies:

• Japanese Yen (JPY): ~0.1% rate → borrow nearly free
• Swiss Franc (CHF): ~1.5% rate → low funding cost
• Euro (EUR): ~3.5% rate → below USD, can fund at discount

Annual expected carry (2026, before price changes):

• Equal-weight long/short portfolio: ~5-7% annual carry income
• Currency moves can add or subtract 3-10% → total return highly variable

The Catastrophic Carry Crashes

Understanding carry crashes is essential. They are violent, fast, and highly correlated across all carry positions simultaneously:

The pattern: Carry crashes are "picking up nickels in front of a steamroller." You earn steady carry income for months/years, then lose it all in a few weeks during a global risk-off event. The strategy is profitable over long periods, but the drawdowns are stomach-churning.

Part 2: Bond Carry — Riding the Yield Curve

Roll-Down: The Other Bond Return

Bond investors focus on yield as the primary return. But there's a second return source: "roll-down." As a bond ages from 10-year to 9-year maturity, it moves down the yield curve—and since the curve typically slopes upward, a shorter-maturity bond has a lower yield (higher price).

Example: U.S. Treasury yield curve (hypothetical 2026):

• 10-year yield: 4.5%
• 9-year yield: 4.3%
• Roll-down: If you buy the 10-year today and hold for 1 year, it becomes a 9-year bond
• Price appreciation from roll: (4.5% - 4.3%) × duration (~8.5) = +1.7% price gain
• Total return: 4.5% yield + 1.7% roll-down = 6.2% in a flat yield-curve environment

This is bond carry: the expected return from holding a bond even if yields don't change.

Institutional Bond Carry Trade

Goldman Sachs Fixed Income and PIMCO use the following structure for bond carry:

Long: The steepest part of the yield curve (10-year to 30-year) where roll-down is highest

Hedge: Duration risk via interest rate futures (5-year Treasury futures) to remove the interest rate risk

Net exposure: Roll-down premium with near-zero duration risk

Expected carry (normal curve): 1.5-2.5% annually above short-term rates

Bond Carry Implementation for Retail

Pure bond carry is difficult to implement without futures. Practical alternatives:

• Long-duration bond ETF + short-duration hedge: Buy TLT (20+ year), short IEF (7-10 year) → net roll-down exposure without full duration risk
• PIMCO Active Bond ETF (BOND): PIMCO's managers actively harvest roll-down across the curve
• PIMCO Enhanced Short Maturity (MINT): Positive carry short-duration exposure
• Yield curve steepener position: If curve is flat/inverted, bond carry is poor—wait for normalization

Part 3: Commodity Carry — The Roll Yield

Contango vs. Backwardation

Commodity futures markets produce carry through their "term structure"—whether futures prices are above or below spot prices:

Backwardation (positive carry):

• Spot price > futures price (e.g., crude oil at $80 spot, $76 for 3-month futures)
• Owning the futures: You sell at $76 per contract initially, then the futures price converges to $80 at expiry → +$4 gain
• This "roll yield" is earned regardless of what happens to the spot price
• Commodities in backwardation 2024: Crude oil, natural gas (seasonal), cocoa

Contango (negative carry):

• Spot price < futures price (e.g., gold at $2,000 spot, $2,080 for 3-month futures)
• Owning futures: You sell at $2,080 initially, futures falls to $2,000 at expiry → -$80 loss from roll
• This is why commodity ETFs like USO (oil) dramatically underperformed spot oil: roll yield was deeply negative
• USO lost 50% of NAV from 2009-2015 while oil was flat—entirely from negative roll yield

Commodity Carry Strategy: Long Backwardation, Short Contango

The strategy:

• Rank 20+ commodity futures markets by roll yield (steepest backwardation to deepest contango)
• Long top quartile (most backwardated): crude, gasoline, coffee, cocoa
• Short bottom quartile (most contango): gold, silver, natural gas (when in contango)
• Rebalance monthly

Historical performance (2000-2024):

• Annual return: +6.8%
• Volatility: 14.2%
• Sharpe: 0.48
• Correlation to S&P 500: +0.12
• Best years: 2007-2008 (+28%), 2021-2022 (+35%) — inflation environments

ETF proxy: PDBC (Invesco Commodity Strategy) uses optimized roll methodology (not just nearest-month roll). Better than DJP or GSG which have severe roll drag.

Part 4: Equity Carry — Dividends and Buybacks

Total Shareholder Yield: The Right Measure

Standard dividend yield misses share buybacks, which are economically equivalent to dividends. Total shareholder yield (dividend yield + buyback yield) is the correct carry measure for equities.

2024 S&P 500 Total Shareholder Yield:

• Dividend yield: 1.3%
• Net buyback yield: 2.1%
• Total shareholder yield: 3.4%

Equity carry strategy: Long stocks with high total shareholder yield (above 5%), short stocks with near-zero or negative yield (companies issuing shares).

Performance (1990-2024, Research Affiliates data):

• Long high-yield quintile vs. short low-yield quintile: +3.8% annually
• Correlation to market: -0.05 (nearly zero)
• Best environment: Low interest rates (high-yield equities substitute for bonds) or inflationary environments
• ETF access: SPHD (Invesco S&P 500 High Div Low Vol), VYM (Vanguard High Dividend Yield), SCHD (Schwab Dividend)

Part 5: Volatility Carry — Harvesting the VRP

The Most Robust Carry Premium

We covered this in depth in the Systematic Options Income article. In the carry framework, volatility carry is simply: sell implied volatility (high carry), buy realized volatility (low carry).

Why vol carry belongs in a carry portfolio:

• Correlation to FX carry: +0.15 (low, but positive)
• Correlation to equity carry: +0.08
• Best when FX carry fails (stress events): Vol carry is highest (VIX spikes → more premium)
• They partially hedge each other: Carry crashes (FX/EM) → VIX spikes → vol selling premiums increase

Implementation:

• Covered calls on SPY (simple, retail-friendly)
• VXX short (complex, significant tail risk—not recommended without education)
• JEPI/JEPQ (managed fund approach)

Part 6: Multi-Asset Carry Portfolio Construction

AQR's Approach: Risk-Parity Carry

AQR's carry fund (AQCIX) doesn't equal-weight carry across asset classes—it risk-parity weights them. Each carry type contributes equal risk (volatility) to the portfolio.

Why risk parity for carry allocation?

• FX carry is more volatile than bond carry (appropriate to weight FX lower)
• EM carry is more volatile than G10 carry (EM gets smaller allocation)
• Equal risk contribution → no single carry type dominates performance
• AQR research shows risk-parity carry Sharpe ~0.8 vs. ~0.5 for equal-weight carry

Practical implementation for individual investors:

Blended expected carry: ~3-5% annually above cash rate, with ~8-10% total volatility

Part 7: The Critical Risk Management Layer

Why Unmanaged Carry Blows Up

The history of carry failures reveals a pattern: investors with too much carry, too much leverage, no crash hedges. The most famous examples:

LTCM (1998): Massive fixed income carry trades (convergence plays). Russian default → correlations went to 1.0, all positions lost simultaneously. $4.6B loss in 6 weeks.

Carry funds in 2008: G10 carry funds lost 25-35% in October-November alone. Most had no crisis hedge.

LJM Partners (2018): VIX carry (short volatility) blew up in "Volmageddon" on February 5, 2018. VIX spiked from 17 to 37 in one day. LJM lost 80%, fund closed.

Institutional Risk Management for Carry

Rule 1: Size carry positions at 1/3 to 1/2 of what feels right

• The inevitable carry crash requires you to hold through drawdown
• If 10% carry position loses 30%, that's -3% portfolio impact—tolerable
• If 30% carry position loses 30%, that's -9% portfolio impact—potentially fatal

Rule 2: Diversify across carry types

• FX and vol carry are both hurt by risk-off events but at different times and magnitudes
• Commodity carry often gains during global risk-off if driven by supply shock
• The 2022 example: FX carry suffered from USD strength (-8%), but commodity carry exploded (+35%) on energy crisis. Net carry portfolio: +12%

Rule 3: Include crash hedges

• Buy put options on the S&P 500 (1-2% of portfolio per year)
• Carry crashes correlate with equity crashes → equity puts provide partial carry hedge
• Alternatively: Maintain managed futures allocation (trend-following profits from carry crashes)

Rule 4: Momentum filter

• If a carry position starts losing (price decline erodes carry income), close the position
• Classic mistake: Hold Turkish lira as it falls from TRY8 to TRY20, "averaging down" because carry is now 25%
• AQR's carry research: Adding 12-month momentum filter improves carry Sharpe from 0.55 to 0.74

import pandas as pd
import numpy as np
import yfinance as yf

def compute_carry_portfolio(assets, carry_rates, lookback=252):
    """
    Build a risk-parity carry portfolio with momentum filter.

    assets: list of ticker symbols
    carry_rates: dict {ticker: annual_carry_rate}
    lookback: days for momentum filter (default 12 months)
    """
    # Download price data
    prices = yf.download(assets, period='2y')['Adj Close']
    returns = prices.pct_change().dropna()

    # Momentum filter: only include positions with positive 12-month return
    momentum_12m = prices.iloc[-1] / prices.iloc[-lookback] - 1
    momentum_filter = momentum_12m > 0

    # Carry scores (only include if momentum is positive)
    carry_scores = {}
    for asset in assets:
        if momentum_filter.get(asset, True):  # Default to include if no momentum data
            carry_scores[asset] = carry_rates.get(asset, 0)
        else:
            print(f"  ⚠️  {asset}: Momentum filter triggered (12m return: {momentum_12m.get(asset, 0):.1%})")
            carry_scores[asset] = 0  # Exclude from portfolio

    # Risk-parity weights: inverse volatility of each asset
    vol = returns.std() * np.sqrt(252)  # Annualized vol
    inv_vol = 1 / vol

    # Normalize weights (only for assets passing momentum filter)
    active_assets = [a for a in assets if carry_scores.get(a, 0) > 0]
    if not active_assets:
        print("No assets passed momentum filter. Holding cash.")
        return {}

    total_inv_vol = sum(inv_vol[a] for a in active_assets)
    weights = {a: inv_vol[a] / total_inv_vol for a in active_assets}

    # Calculate expected portfolio carry
    portfolio_carry = sum(weights[a] * carry_scores[a] for a in active_assets)

    print(f"\nCARRY PORTFOLIO ALLOCATION:")
    print(f"{'Asset':<10} {'Weight':>8} {'Carry':>8} {'12m Mom':>10}")
    print("-" * 40)
    for asset in active_assets:
        mom = momentum_12m.get(asset, 0)
        print(f"{asset:<10} {weights[asset]:>8.1%} {carry_scores[asset]:>8.1%} {mom:>10.1%}")

    print(f"\nPortfolio Expected Carry: {portfolio_carry:.1%}/year")
    print(f"Active positions: {len(active_assets)}/{len(assets)}")

    return weights

# Example: Multi-asset carry portfolio
assets = ['FXA',  # AUD (high carry currency)
          'TLT',  # Long bond (roll-down carry)
          'SCHD', # High dividend equities
          'PDBC', # Commodity carry
          'JEPI'] # Volatility carry

carry_rates = {
    'FXA':  0.038,  # AUD-USD carry ~3.8%
    'TLT':  0.022,  # Roll-down ~2.2%
    'SCHD': 0.034,  # Dividend yield ~3.4%
    'PDBC': 0.048,  # Commodity roll ~4.8%
    'JEPI': 0.095,  # Vol carry ~9.5%
}

weights = compute_carry_portfolio(assets, carry_rates)
                    

Part 8: Carry in a FIRE Portfolio Context

The Income-Generation Role

Carry strategies are natural complements to a FIRE withdrawal strategy:

Standard withdrawal problem: In a down market, you must sell shares to fund expenses—triggering sequence of returns risk.

Carry solution: Carry income is largely independent of price appreciation. Even if equity prices fall, carry positions generate income (dividend yield, options premium, roll yield) that can fund expenses without share sales.

Case study: $1.5M FIRE portfolio during 2022 bear market

The carry-augmented portfolio sold 80% fewer shares during a bear market—dramatically reducing sequence of returns risk while generating nearly equivalent total income.

Conclusion: Carry as a Systematic Income Source

Carry is not a magic free lunch. It's compensation for bearing specific risks: currency crashes, credit events, volatility spikes. But with proper diversification across carry types (FX, bonds, equity, commodity, vol) and strict risk management (momentum filter, crash hedges, appropriate sizing), carry can generate 3-6% annual income with low equity correlation.

For FIRE investors, the most practical carry implementation is:

1. High-dividend equity ETFs (SCHD, VYM) — equity carry, low complexity
2. JEPI or systematic covered calls — volatility carry, monthly income
3. PDBC — commodity carry, inflation hedge
4. PIMCO BOND — bond carry, professional management
5. Total allocation: 20-30% of portfolio, generating 4-7% income annually

Further Reading

Research:

• Asness, C. et al. (2013): "Carry" — NBER Working Paper 19347
• Lustig, H. & Verdelhan, A. (2007): "The Cross Section of Foreign Currency Risk Premia"—American Economic Review
• AQR: "Understanding Style Premia" (2014) — free download at aqr.com

Related Articles:

• Alternative Risk Premia (broader framework)
• Systematic Options Income (vol carry)
• Managed Futures (trend-following hedges carry crashes)
• Private Credit (credit carry)