Asset classes · Commodities one by one
Commodity indexes: S&P GSCI, BCOM and DBC’s index compared
How the S&P GSCI, the Bloomberg Commodity Index and the index behind DBC choose weights and roll futures, and how spot, excess and total return versions differ.
On this page
"Commodities rose 10% this year" sounds like a fact, but it depends entirely on which commodities, in what proportions, held through which futures contracts, and whether the interest on the cash behind them is counted. A commodity index answers those questions with rules. Different indexes answer them differently, and the answers matter more than for stock indexes: two broad commodity benchmarks can differ by many percentage points in a single year.
This lesson compares the three benchmarks an American investor meets most often: the S&P GSCI, the Bloomberg Commodity Index (BCOM), and the DBIQ Optimum Yield Diversified Commodity Index behind the DBC fund. It shows how their weights and roll rules differ, and explains the three versions every commodity index comes in, spot, excess return and total return, with a worked example using the actual interest rate on Treasury bills. The mechanics of contango and roll yield are in Contango, backwardation and roll yield: why USO lags oil; fund legal forms and taxes are in Commodity ETFs: gold trusts, futures pools and their tax forms.
What an index has to decide
A stock index can simply hold shares. A commodity index cannot hold barrels or bushels, so it is a set of rules about futures contracts:
- Which commodities. Usually the most liquid US and London contracts: energy, industrial and precious metals, grains, softs and livestock.
- How much of each. By world production, by trading volume, by fixed weights, or a mix, with or without caps.
- Which contract month. The nearest one, or one further out.
- When to roll. Futures expire, so the index must sell the expiring contract and buy a later one on a set schedule.
- What counts as return. Price change only, price plus roll, or price plus roll plus the interest on the cash that collateralizes the position.
Every one of these choices changes the result. Weights decide whether a year of falling oil and rising gold is a gain or a loss; the roll decides how much contango costs.
Three benchmarks
S&P GSCI: production-weighted
The S&P GSCI, started by Goldman Sachs in 1991 and now run by S&P Dow Jones Indices, sets each commodity's weight from a five-year average of world production, with a check on trading volume. The idea is economic: the index should look like the world's commodity output by value. Because the world produces far more oil, by value, than anything else, energy dominates. S&P's November 2026 announcement gives 2026 reference weights of 51.78% for energy and 17.01% for agriculture, 11.19% for livestock, 12.13% for industrial metals and 7.89% for precious metals; for 2027 energy stays the largest at 51.16%. The index generally holds the nearest liquid contract and rolls over five business days early in the month, per its methodology.
The consequence: the S&P GSCI behaves much like an oil index with extras. It did very well when oil soared, and very badly when it collapsed.
BCOM: liquidity-weighted with caps
The Bloomberg Commodity Index, formerly the Dow Jones-UBS index, sets target weights each year using trading liquidity and production in a 2:1 ratio, then applies diversification caps. In its 2027 target weights announcement Bloomberg lists them:
- no group (energy, grains, industrial metals and so on) above 33%,
- no single commodity together with its derivatives above 25%,
- no single commodity above 15%,
- no commodity below 2%, as liquidity allows.
The result is far more balanced. For 2026 BCOM's weights were 29.44% energy, 21.15% grains, 9.17% softs, 5.64% livestock, 15.76% industrial metals and 18.84% precious metals, with gold alone at 14.90%. Gold, small in world production by value but heavily traded, gets nearly twice the weight it has in the GSCI's whole precious-metals sector.
DBC's index: optimum yield
The Invesco DB Commodity Index Tracking Fund (DBC) follows the DBIQ Optimum Yield Diversified Commodity Index Excess Return. For most of its life the index held 14 commodities whose weights were reset each November to fixed base proportions. The fund's 10-K for 2018 lists them at the end of that year: four oil contracts and natural gas (49.6% together), gold and silver (11.9%), aluminum, zinc and copper (13.7%), and corn, wheat, soybeans and sugar (24.8%), with no livestock.
Its distinctive rule is the roll. Instead of always buying the next contract, the index looks at eligible contracts up to the thirteenth month ahead and picks the one with the most favorable implied roll yield: in contango, the one whose price is least above the expiring contract per month of distance; in backwardation, the one that gains most. It is designed to soften contango's cost, not to escape it.
The index has since been broadened. The fund's November 2025 prospectus lists 28 index commodities, including livestock, with weights based on an average of liquidity and production, no sector above 40%, no single commodity above 12.5%, and an annual rebalance on the sixth business day of November. The fund charges a 0.85% management fee.
Spot, excess return and total return
Every commodity index is published in up to three versions. They answer different questions.
- Spot index. Tracks the prices of the futures the index holds, without accounting for rolling. It is a price gauge, like the front-month charts in the single-commodity lessons, and no one can earn it.
- Excess return (ER). Tracks what a fully collateralized futures position earns from the futures alone, including the gain or loss at each roll. It is the return above cash, which is why it is called "excess". DBC's index is an ER index.
- Total return (TR). The excess return plus the interest earned on the collateral, conventionally the 3-month Treasury bill rate. This is the closest to what a fund that holds futures and keeps its cash in T-bills could earn before fees.
The figure shows why the versions diverged so much over two decades. From 2009 to 2015 T-bills earned at most 0.15% a year, so total and excess return indexes were almost the same. In 2006–2007 and 2023–2024 the collateral added 4.4% to 5.2% a year. Compounded from March 2006 to September 2026, T-bills grew by about 42%. Over almost the same span, March 2006 to October 2026, DBC's share price rose 41%, from $23.94 to $33.72, a fund whose price includes interest it retains and is net of fees. Put simply, two decades of a broad commodity fund roughly matched cash in price, with a ride that included a 72% fall from June 2008 to February 2016.
Pitfalls and limitations
- "Commodities" is a weighting choice. An energy-heavy index and a capped one can give opposite answers in the same year. Check the sector weights before reading a headline.
- Know which version is quoted. A total return index can beat its excess return sibling by several points a year when rates are high. Comparing a TR index with an ER fund, or a spot index with either, is comparing different things.
- Weights change. BCOM and the GSCI reset every year, and DBC's index went from 14 to 28 commodities. A long backtest may describe a different basket from today's.
- Optimum yield is not free. Choosing deferred contracts reduces roll costs in contango but changes the exposure: a contract a year out moves less with today's news.
- Price-only data understate fund returns when a fund pays out its interest as distributions. The DBC figures here are share prices only.
The three indexes side by side
| Feature | S&P GSCI | BCOM | DBC's index (DBIQ OY) |
|---|---|---|---|
| Weighting | World production (5-yr avg) | Liquidity 2/3, production 1/3, capped | Liquidity and production, capped |
| Energy (latest) | 51.8% (2026) | 29.4% (2026) | 49.6% (Dec 2018, old 14-commodity version) |
| Caps | None of note | Group 33%, commodity 15% | Sector 40%, commodity 12.5% (2025) |
| Roll | Nearby contract, set window | Set window, fixed contract schedule | Best implied roll yield, up to 13 months |
How Finabase does it
Finabase does not publish or chart commodity index levels; the S&P GSCI and BCOM are licensed index data. What the app does have are the parts and one of the funds. The Futures page lists the front-month contracts these indexes are built from, energy, metals, grains and softs, with returns from one week to two years. DBC has an ordinary ETF symbol page with its price history. The Treasury page shows today's Treasury yields from one month out, which is the current rate on the collateral that turns an excess return into a total return. This lesson's weights come from the index sponsors' documents and its T-bill returns from the Federal Reserve's series on FRED.
Frequently asked questions
What is the difference between the S&P GSCI and the Bloomberg Commodity Index?
The S&P GSCI weights commodities by world production, which makes it about half energy. BCOM weights mostly by trading liquidity and caps any group at 33% and any commodity at 15%, which gives energy about 29% and gold about 15% in 2026.
What is a total return commodity index?
It is the excess return from holding and rolling futures plus the interest earned on the collateral, usually 3-month T-bills. In 2023 that interest added 5.19%; in 2021 it added 0.04%.
What index does DBC track?
The DBIQ Optimum Yield Diversified Commodity Index Excess Return. Its roll picks, for each commodity, the contract up to 13 months out with the best implied roll yield. Its November 2025 prospectus lists 28 commodities.
Why do commodity indexes give different returns?
Their weights differ (energy is about 52% of the GSCI and 29% of BCOM), their roll rules differ, and some versions include T-bill interest while others do not.
Summary
A commodity index is a set of rules about futures: which commodities, how much of each, which contracts and when to roll them, and whether to count the interest on collateral. The S&P GSCI's production weights make it about half energy; BCOM's liquidity weights and caps spread it out, with gold its largest single holding; DBC's index picks contracts by roll yield and has grown from 14 to 28 commodities. Spot, excess return and total return versions differ by roll and by T-bill interest, which ranged from 0.04% to 5.19% a year in 2006–2026. Before quoting "commodities returned X", say which index and which version.
Check your understanding
4 questions on this lesson. Each answer comes with the reason, and nothing is scored.
In a year when oil falls 30% and gold rises 30%, which index would you expect to do better, and why?
- The S&P GSCI, because it is production-weighted
- BCOM, because energy is capped near 30% and gold alone is about 15%, while the GSCI is about half energy
- They would return the same
Show the answer
B. BCOM, because energy is capped near 30% and gold alone is about 15%, while the GSCI is about half energy Weights decide the result. With about 52% in energy and under 8% in precious metals, the GSCI is hit much harder by an oil fall and gains less from gold.
An excess-return commodity index ends the year down 2%. The 3-month T-bill rate that year compounded to 5%. What did the total-return version earn?
- −2%
- About +2.9%
- +7%
Show the answer
B. About +2.9% Total return multiplies the two: 0.98 × 1.05 − 1 = 2.9%. The interest on the collateral more than offset the futures loss.
What does the "optimum yield" roll in DBC’s index try to do?
- Avoid holding futures at all
- Pick, for each commodity, the contract up to 13 months out with the most favorable implied roll yield, to reduce contango costs
- Roll only when prices are rising
Show the answer
B. Pick, for each commodity, the contract up to 13 months out with the most favorable implied roll yield, to reduce contango costs Instead of always buying the next month, it compares eligible contracts and chooses the one whose price structure costs least to hold. It softens contango but changes the exposure.
Why were total-return and excess-return commodity indexes almost identical from 2009 to 2015?
- Commodity prices did not move
- T-bills earned at most 0.15% a year, so the collateral added almost nothing
- The index sponsors suspended the total-return versions
Show the answer
B. T-bills earned at most 0.15% a year, so the collateral added almost nothing The only difference between the two versions is the interest on collateral. With 3-month T-bill rates near zero, there was little to add.
Exercise
One task to do in Finabase. Nothing is graded or saved.
Open the DBC chart at the longest range and find its 2008 peak, its 2020 low and where it is today. Then open the Treasury page and note today’s 1-month and 3-month bill yields. Roughly how much a year would the collateral behind a commodity index earn at today’s rates, compared with 2010–2015?
Open the DBC chartWhat you should find
DBC fell about three-quarters from mid-2008 to April 2020 before recovering. Today’s bill yields are what a total-return index adds to the futures result each year; in 2010–2015 that was close to zero, so the difference between index versions changes a lot with interest rates.
Sources and further reading
- S&P Dow Jones Indices — S&P GSCI methodology
- S&P Dow Jones Indices — S&P GSCI 2027 contract production weights announcement, November 6, 2026 (2026 and 2027 reference percentage dollar weights)
- Bloomberg — Bloomberg Commodity Index (BCOM) methodology, September 2026
- Bloomberg — BCOM 2027 target weights announcement (with 2026 weights and diversification caps), October 29, 2026
- Invesco DB Commodity Index Tracking Fund — Form 10-K for 2018 (index commodities, fund weights, Optimum Yield roll)
- Invesco DB Commodity Index Tracking Fund — prospectus, November 2025 (28 index commodities, caps, annual rebalance, 0.85% fee)
- Federal Reserve Board — 3-month Treasury bill secondary market rate (FRED series TB3MS), downloaded 2026-11-18
- Finabase chart data — DBC monthly closes, March 2006 to October 29, 2026 (price only)
