Guide

Roll yield, contango and backwardation explained

Harbor Capital's 8% strategic commodities sleeve tracked spot WTI within 40 basis points for three years — then lost 11% in six months while crude was flat. The culprit was not bad directional calls; it was roll yield drag. The fund held front-month oil futures through a steep contango curve: each monthly roll sold an expiring contract near $78 and bought the next month near $81. Spot did not move, but the position bled 3% per roll. Risk managers refactored the sleeve to blend second-month contracts when the one-month calendar spread exceeded a z-score threshold, and to allocate part of the book to backwardation-friendly markets (copper, live cattle) when the aggregate GSCI curve slope turned positive. Tracking error to spot fell; roll drag over the next year was cut by roughly two-thirds.

Roll yield is the return from moving a futures position from a near-dated contract to a farther-dated one as expiration approaches. In contango, farther months trade above nearer months — rolling costs money. In backwardation, farther months trade below — rolling earns carry. This guide covers futures curve anatomy, calendar spread math, how commodity ETFs inherit roll decay, index and Treasury curve nuances, the Harbor Capital commodities sleeve refactor, a technique decision table against spot and perpetual swaps, pitfalls, and a production checklist.

The futures curve and why shape matters

A futures curve plots contract prices by delivery month. Each point is a tradable obligation to buy or sell a standardized quantity on a fixed date. Curves are not predictions of spot — they embed storage costs, financing, convenience yield, and supply-demand imbalances at each horizon.

Contango

Contango means the curve slopes upward: deferred contracts trade at higher prices than nearby ones. Physically storable commodities often sit in contango when inventory is ample and storage is cheap relative to immediate demand. Financial futures can contango when the risk-free rate and dividend or yield adjustments push fair value above spot (cost-of-carry model).

Backwardation

Backwardation is the opposite slope: nearby contracts trade above deferred. It appears when immediate supply is tight, storage is scarce, or holders of physical inventory earn a convenience yield from having barrels or bushels on hand today. Energy markets spike into backwardation during geopolitical shocks; agricultural curves backwardate ahead of harvest when old-crop stocks run low.

Whether you hold futures for one month or ten years, you eventually roll — sell the contract nearing expiry and buy the next liquid month. The price difference between those two legs is roll yield (positive in backwardation, negative in contango).

Roll yield math: calendar spreads

Define the calendar spread between month t+1 and month t as:

Spread = Ft+1 − Ft

A long position in the front month that rolls into the second month effectively sells Ft and buys Ft+1. Roll yield over that roll is approximately −Spread / Ft as a percentage (ignoring bid-ask and execution timing).

Worked example: WTI in contango

  • Front-month future: $78.00/bbl
  • Second-month future: $81.00/bbl
  • Spread: +$3.00 (contango)
  • Roll yield: roughly −3.85% on notional for that single roll

If spot WTI stays at $78 through the roll, the futures investor still loses ~3.85% on the position change. Twelve monthly rolls at similar slope compound into severe underperformance versus spot — the pattern that damaged crude ETFs in prolonged contango regimes.

Worked example: copper in backwardation

  • Front month: $4.20/lb
  • Second month: $4.10/lb
  • Spread: −$0.10 (backwardation)
  • Roll yield: roughly +2.38% for the roll

With spot unchanged, the long futures holder gains from rolling down the curve. This is the commodity analogue of positive carry in FX — earned without directional spot movement.

Annualized roll yield

Practitioners often annualize one-month roll yield by multiplying by roll frequency (12 for monthly commodities, 4 for quarterly index rolls). Use this only as a rough gauge: curve shape changes every month, and rolls are not always exactly 30 days apart. For attribution, log each realized roll return rather than extrapolating a single spread snapshot.

How commodity ETFs inherit roll decay

Exchange-traded products that track commodities via futures — not physical storage — must roll before expiry. Their return decomposes into:

  • Spot (or index) return — movement in the underlying cash market
  • Roll yield — curve shape at each roll date
  • Collateral yield — interest on cash margin (often modest)
  • Tracking error — fees, slippage, and index methodology quirks

Single-commodity ETFs concentrated in front-month contracts are most exposed. When storage costs dominate (oil in floating tanks, natural gas at hub), contango can persist for years. Broad indices like the S&P GSCI roll across multiple commodities on fixed calendar windows (typically business days 5–9 and 10–14 each month), which front-runs some roll flow and changes the effective spread paid versus a naive last-day roll.

Optimized roll products attempt to minimize contango by selecting farther months with flatter curves when the front spread exceeds a threshold. The trade-off is lower liquidity and different exposure to spot shocks in deferred contracts. No roll rule eliminates curve risk — it reallocates it.

Beyond commodities: equities, rates and crypto

Equity index futures

S&P 500 and Nasdaq futures typically trade at a premium to spot reflecting financing minus expected dividends (fair value). Long index futures earn negative roll yield versus spot in calm markets (equity contango). Short futures earn positive roll carry — one reason cash-and-carry arbitrage desks sell futures against the basket.

Treasury futures

Bond futures curves embed repo rates, conversion factors, and cheapest-to-deliver dynamics. Roll yield interacts with duration and rate sensitivity; attribution requires contract-specific CTD math, not a single spread rule.

Perpetual swaps vs dated futures

Perpetual futures avoid calendar rolls but charge periodic funding tied to the basis between perp and spot. Funding is the continuous analogue of roll yield: positive funding (longs pay shorts) resembles contango drag; negative funding resembles backwardation carry. Compare dated-roll strategies and perps on total all-in carry, not spot beta alone.

Harbor Capital commodities sleeve refactor

Before the refactor, Harbor's passive sleeve held front-month weights matching the Bloomberg Commodity Index. Risk reported roll drag separately but did not act on it. After the 11% flat-crude drawdown, the team implemented three rules:

  1. Spread z-score gate — when the one-month WTI calendar spread exceeded +1.5σ vs a five-year rolling window, shift 50% of the oil allocation to month-two contracts until the spread normalized.
  2. Cross-commodity tilt — overweight commodities in backwardation (copper, cattle) and underweight the worst contango offenders (nat gas, coffee) within a ±2% active risk budget versus the benchmark.
  3. Roll window awareness — avoid initiating large positions in GSCI roll days 5–9 when liquidity demand widens spreads; use TWAP execution across the roll window.

The sleeve still diverged from spot by design — the goal was to separate intentional roll exposure from accidental drag. Monthly attribution now reports spot return, roll return, and collateral yield on every commodity line.

Technique decision table

Goal Prefer Avoid when
Pure spot beta, no curve risk Physical ETF (gold bars), spot crypto, equity shares Storage/insurance costs prohibitive (crude, nat gas)
Leveraged commodity exposure, accept roll Front-month futures or single-commodity ETP Steep contango (check 1-month spread z-score)
Reduce contango drag Optimized-roll ETP, month-two+ when spread wide Need tight spot tracking in backwardation spikes
Earn carry without dated rolls Perpetual swap (short if funding positive) Funding regime flips; exchange/counterparty risk
Diversified commodity beta Broad index (GSCI, BCOM) with documented roll rules Ignoring index roll calendar front-running
Hedge physical inventory Short nearby futures (earn contango if hedger) Basis risk between local cash and benchmark future
Macro carry expression FX carry or backwardated commodity longs Risk-off squeezes flatten curves quickly

Common pitfalls

  • Equating ETF return to spot — years of contango can make oil ETPs lose while spot rallies; read the prospectus roll methodology.
  • Ignoring roll calendar — index funds cluster rolls on known days; spreads widen for everyone, not just you.
  • Using spot charts to manage futures — P&L includes roll; mark positions on futures curves, not cash alone.
  • Chasing last month's backwardation — curves mean-revert; cattle backwardation does not guarantee next month's roll yield.
  • Confusing fair-value contango with stress backwardation — equity index contango is often normal; commodity spikes need different sizing.
  • Single-spread extrapolation — annualizing one fat spread overstates expected drag; use distributions.
  • Deferred-month liquidity — month-six oil may look flat but costs more to trade; slippage eats theoretical roll savings.
  • Perp funding vs dated roll — switching instruments changes carry mechanics and tail risks; compare all-in costs.

Production checklist

  • Plot front two months per commodity; flag contango/backwardation daily.
  • Compute one-month calendar spread and z-score vs rolling history.
  • Decompose backtest returns into spot, roll, and collateral components.
  • Document index roll windows (e.g., GSCI days 5–9 and 10–14).
  • Stress-test sleeves for 2008–2009 oil contango and 2020 WTI negative expiry.
  • Set spread gates before shifting to deferred contracts; log override reasons.
  • Measure execution slippage on rolls separately from theoretical roll yield.
  • Compare perp funding vs dated roll when choosing crypto commodity proxies.
  • Report roll attribution monthly to LPs with commodity line items.
  • Review roll rules when storage capacity or shipping costs structurally shift.

Key takeaways

  • Roll yield is the P&L from moving futures to the next contract — negative in contango, positive in backwardation.
  • Spot and futures total returns diverge whenever the curve is not flat; commodity ETFs can bleed for years in steep contango.
  • Calendar spread math is simple, but execution timing, index roll windows, and liquidity matter as much as the quoted spread.
  • Harbor Capital cut roll drag by gating wide spreads, tilting toward backwardated markets, and attributing spot vs roll explicitly.
  • Perpetual funding is the continuous cousin of roll yield — compare all-in carry when choosing instruments.

Related reading