Carry & Mean Reversion in Futures
A comprehensive architectural review of systematic carry and counter-trend strategies for futures markets. Strictly curve-based roll yield and statistical mean reversion — daily-close, low-frequency implementation. FX interest rate differentials are explicitly excluded.
1. Commodity Futures Carry — Core Methodology
This section explores the theoretical and practical underpinnings of commodity curve carry. We define the mechanics of roll yield, the economic rationale for its persistence, and analyze which sectors historically provide the most robust signals.
Definition & Calculation
In systematic futures trading, Carry (or Roll Yield) is the return generated by a futures contract converging toward the spot price over time, assuming the curve shape remains static. Practitioners generally calculate annualized carry using the front and second nearby contracts:
- Backwardation (Positive Carry): Front > Second. The deferred contract "rolls up" the curve toward the higher front price.
- Contango (Negative Carry): Front < Second. The deferred contract "rolls down" toward the lower front price.
Theoretical Basis
Keynes' Normal Backwardation: Producers hedge by selling futures, driving deferred prices below expected spot. The positive carry is the risk premium paid to speculators for providing liquidity.
Theory of Storage & Convenience Yield: Deferred price = spot + cost of carry (storage, insurance, financing) − convenience yield. High convenience yield during shortages drives backwardation.
Signal Construction & Seasonality
To avoid seasonal contamination (e.g., old crop vs. new crop corn spreads), robust models use spot-adjusted carry or compare contracts exactly one year apart to isolate structural carry from calendar patterns.
Predictive edge often increases by measuring the slope of the entire liquid term structure (regression across the first 4–6 contracts) rather than just front/second ratio.
Interactive Term Structure Dynamics
Toggle below to visualize how the passage of time generates roll yield for a static position.
In Backwardation, a long deferred position yields positive carry as its price rolls up to spot over time.
Historical Carry Persistence by Sector
Energy and Base Metals have historically exhibited the most robust carry premia due to high storage costs, supply shocks, and distinct producer hedging pressures.
Illustrative Sharpe estimates. Not drawn from a single study.
2. Carry Strategies in Other Asset Classes
Carry is a ubiquitous cross-asset phenomenon. Expanding beyond commodities provides significant diversification benefits, as the macroeconomic drivers of carry vary by asset class.
📈 Equity Index Futures
Calculation: Expected Dividend Yield minus Risk-Free Financing Cost.
If the dividend yield exceeds the financing rate, the futures contract trades at a discount. Holding long generates positive carry as the discount decays. Historically strong but highly correlated to the underlying equity risk premium.
💵 Fixed Income Futures
Calculation: Yield curve roll-down + cheapest-to-deliver (CTD) dynamics.
In a normal upward-sloping curve, a bond's yield decreases as it ages, increasing its price. Provides excellent diversification to commodity carry; suffers during aggressive central bank hiking cycles.
3. Practical Implementation Requirements
The following rules define the systematic boundaries of the strategies discussed in this report.
4. Counter-Trend & Mean Reversion
While Trend and Carry rely on persistence, Counter-Trend seeks to profit from the elastic nature of markets overreacting to short-term events — fading statistically extreme positioning or pricing anomalies.
Fading Curve Extremes
The most robust carry-based counter-trend strategy involves fading extreme backwardation or extreme contango. When a market enters massive backwardation (e.g., agricultural supply shock), speculators crowd the long side. If the structural shortage resolves, the curve violently collapses back to normal.
Common Failure Modes
- "Left-Tail" Structural Shifts: Fading extreme backwardation fails catastrophically if a genuine, unresolvable supply shortage occurs (e.g., LME Nickel 2022). The curve goes asymptotic.
- Whipsaws: Fading price extremes during a low-volatility grind leads to multiple small losses before mean reversion occurs.
Diversification with Trend
Counter-trend is the theoretical antidote to the "trend-following smile." It performs best in range-bound, high-volatility environments — exactly when trend suffers most. Its inclusion dramatically smooths portfolio equity curves.
5. Combining Carry and Trend
Carry and Time-Series Momentum (Trend) are historically the two most robust risk premia in futures markets. Integrating them creates a synergistic portfolio.
Signal Independence
Literature confirms Carry and Trend exhibit low to slightly negative correlation over the long term (typically −0.1 to +0.1). Trend profits from macro shifts and directional volatility. Carry profits from stability, structural bottlenecks, and risk-premium harvesting in sideways markets.
Practitioner consensus (AQR, Man Group): size these strategies inversely to their volatility — often 50/50 risk parity between Carry and Trend sleeves.
| Correlation | Trend | Carry | Mean Rev. |
|---|---|---|---|
| Trend | 1.00 | 0.05 | −0.35 |
| Carry | 0.05 | 1.00 | 0.12 |
| Mean Rev. | −0.35 | 0.12 | 1.00 |
Stylized long-term estimates across major futures markets.
6. Risk and Regime Considerations
Carry is susceptible to "crash risk" — picking up pennies in front of a steamroller — requiring quantitative regime filters.
When Carry Fails
The Crowded Trade: When capital floods into obvious carry trades, the unwind can be violent — a sudden flattening of the curve against the carry position.
Liquidity Crises: During global margin calls (e.g., March 2020), correlations go to 1 and carry risk premia are rapidly repriced across all asset classes simultaneously.
Regime Filters (Daily Close)
Volatility Scaling: The most robust defense — size positions inversely to trailing realized volatility. As volatility expands, carry exposure is mechanically reduced.
Trend Filter Overlay: Only take the positive carry trade (long) if price is also above its 200-day moving average. Prevents catching falling knives in structurally declining markets.
7. Prioritized Research Library
A curated, actionable reading list focusing on institutional-grade methodologies.
Systematic Trading
BookRobert Carver (2015)
Written by a former AHL quantitative portfolio manager. Provides explicit, mathematically sound frameworks for portfolio construction, position sizing based on instrument volatility, and blending rules (carry + mean reversion). The definitive guide on converting raw signals into a risk-managed portfolio using daily data.
Carry
AcademicKoijen, Moskowitz, Pedersen, Vrugt (2018) — Journal of Financial Economics
The seminal modern paper on cross-asset carry. Formally defines carry calculation for commodity futures and demonstrates its low correlation with trend-following over a multi-decade dataset spanning equities, bonds, commodities, and currencies.
The Strategic and Tactical Value of Commodity Futures
AcademicErb and Harvey (2006) — Financial Analysts Journal
Foundational text establishing that historical equity-like returns of commodity indexes were almost entirely driven by the roll yield of backwardated contracts, not spot price appreciation. Breaks down carry sector by sector.
Commodities for the Long Run
PractitionerLevine, Ooi, Richardson (AQR Capital Management, 2018)
Institutional white paper providing highly practical, implementable rules for combining carry and trend within commodities. AQR demonstrates how a 50/50 blend dramatically increases Sharpe ratio over a long-only commodity benchmark, with specific roll period and contract selection rules.
Demystifying Managed Futures
PractitionerBrian Hurst et al. (AQR White Paper)
Invaluable for its section on non-trend strategies in managed futures. Outlines specific mechanisms and historical thresholds used by institutional CTAs to implement counter-trend/mean-reverting strategies at the daily close.