Futures ETFs Explained: Roll Costs, the USO 2020 Episode and Leveraged Funds

Most commodity ETFs do not own barrels of oil or bushels of corn. They own futures contracts on those commodities and roll them forward before expiry, so their returns depend on futures prices and on the cost or benefit of rolling, not just on the spot price you see in headlines. That is why an oil ETF can lose money over a period when the spot price of oil ends roughly where it started.

In short: A futures-based ETF holds near-dated futures, sells them before they expire, and buys later-dated ones. When the futures curve is in contango (later months more expensive), each roll buys fewer contracts with the same money, which drags on returns. The United States Oil Fund (USO) showed how extreme this can get in April 2020: according to its own filings, it ran out of registered shares, changed which contracts it held several times in two weeks, did a 1-for-8 reverse split, and ended 2020 with its net asset value per share down 67.66%, versus a 20.54% fall in its benchmark futures contract. Leveraged and inverse products add a second effect, daily resetting, that makes long-term returns differ from a simple multiple of the index.

Why ETFs use futures

Holding physical oil or natural gas means paying for storage, insurance, and transport. Futures give price exposure without any of that. A fund puts most of its cash in Treasury bills or similar instruments, posts part of it as margin, and holds futures with a notional value about equal to its assets. The fund earns interest on the cash and the gain or loss on the futures.

Gold is a common exception: many gold ETFs hold physical bullion in vaults, because gold is cheap to store relative to its value.

How rolling works

Futures expire. A fund that wants continuous exposure must roll: sell the contract that is about to expire and buy a later one. Most funds publish a roll schedule. USO’s annual report for 2020 describes a roll period of four days that it extended to ten days starting with the May 2020 roll.

The mechanics of rolling are explained in rolling futures contracts, and the curve shapes that drive roll results are in contango and backwardation.

Worked example: the cost of rolling in contango

This uses the real NYMEX WTI crude oil contract: 1,000 barrels per contract. Prices are illustrative, chosen to resemble the steep contango of spring 2020.

A fund has $100 million invested in the front-month WTI contract at $20.00 per barrel.

Barrels held: $100,000,000 / $20.00 = 5,000,000 barrels (5,000 contracts)

It rolls into the next month, which trades at $25.00.

Barrels after the roll: $100,000,000 / $25.00 = 4,000,000 barrels (4,000 contracts)

The fund still has $100 million, but it now owns 20% fewer barrels of exposure. A month later, suppose nothing has changed in the market: spot oil is still $20, and the contract the fund holds has converged down toward $20 as it nears expiry.

Value: 4,000,000 x $20.00 = $80,000,000

The fund lost 20% even though the spot price did not move. In backwardation the effect reverses: later contracts are cheaper, each roll buys more barrels, and the fund can outperform spot. Real curves are rarely this steep for long, but in April 2020 they were.

The USO episode in April 2020

USO is a commodity pool structured as a limited partnership, listed on NYSE Arca. Its stated objective was to track the daily percentage changes of a benchmark crude oil futures contract, normally the near-month WTI contract (switching to the next month as expiry approached).

What follows is from USO’s own SEC filings: its 8-K current reports from April 2020 and its 2020 annual report on Form 10-K. The background on the oil market itself is in negative oil prices in April 2020.

Running out of shares

Investors poured money into USO as oil prices collapsed in March and April 2020. ETFs create new shares in blocks (“creation baskets”) for authorised participants, but only up to the number of shares registered with the SEC. On April 20, USO disclosed that it had only 299,900,000 registered shares available and might have to suspend creations. On April 21 it issued its remaining registered shares and suspended creation baskets until a new registration statement became effective. Redemptions and exchange trading continued.

With no new shares being created, USO’s market price drifted away from its net asset value. The 10-K reports that on April 21, 2020, USO shares traded at a price 36% above its end-of-day NAV. The premium fell to 8.66% the next day and 1.45% by May 1. USO did not resume offering creation baskets until a new registration was declared effective in June 2020.

Moving away from the front month

USO’s size, its brokers’ limits, and exchange accountability levels pushed it out of the front month:

Date (2020)	What USO’s 8-K filings described
April 16	From April 17, about 80% in the front month and 20% in the second month
April 21	About 40% June, 55% July, 5% a later month
April 22	About 20% June, 50% July, 20% August, 10% September
April 24	About 20% June, 40% July, 20% August, 20% September
April 27	Positions spread from July out to later months, phased in over several days

The 10-K explains why. Letters from CME dated April 16 and April 23, 2020 set an accountability level of 10,000 contracts for the June 2020 contract and maximum long positions for June (15,000), July (78,000), August (50,000), and September (35,000). In May 2020 its futures broker, RBC, told USO it could not hold the June benchmark contract.

USO also announced a 1-for-8 reverse split, effective after the close on April 28, 2020.

The result for 2020

From the 10-K, adjusted for the reverse split:

Measure	Start of 2020	End of 2020	Change
USO NAV per share	$102.27	$33.07	-67.66%
Benchmark WTI futures contract	$61.06	$48.52	-20.54%

USO’s filing says that had it tracked its benchmark exactly, its NAV would have fallen about 42.97% (the benchmark’s 20.54% price fall understates what a fund rolling through the curve would have lost). USO’s actual result was 24.68 percentage points worse than that. The filing explains that during 2020, market conditions and regulatory limits led USO to hold contracts in months other than its benchmark, and that it “has not been able to meet its investment objective as favorably as it has in the past.”

What regulators found

In November 2021, the SEC and the CFTC settled parallel cases. The SEC found that USO and its general partner, United States Commodity Funds, made misleading statements about a limitation imposed by USO’s futures broker: in April 2020 the broker said it would not execute new oil futures positions for USO, and the fund did not fully disclose that limitation until about a month later. USO and its general partner agreed to a $2.5 million penalty, offset by up to $1.25 million paid in the CFTC case; neither admitted nor denied the findings.

Leveraged and inverse funds

Some ETFs aim to return a multiple of a benchmark’s daily return, such as 2x or -1x. Many use futures or swaps to do it. Because they reset exposure every day, their results over longer periods can differ sharply from the multiple of the benchmark’s longer-term return.

Worked example: daily reset

Start with a benchmark at 100. Day 1 it rises 10%; day 2 it falls 10%.

	Day 0	After day 1	After day 2	Two-day result
Benchmark	100	110	99	-1%
2x daily fund	100	120	96	-4%
-1x daily fund	100	90	99	-1%

The 2x fund rose 20% on day 1 and fell 20% on day 2 (from 120, that is 24 points). It lost 4%, not 2 x -1% = -2%. The inverse fund fell 10% then rose 10%, ending down 1% even though the benchmark also fell 1%. In volatile, sideways markets, this decay compounds. Fund prospectuses state these objectives in daily terms for this reason.

A note on taxes and structure

Many US commodity futures funds are structured as partnerships rather than as regular investment companies, and they send investors a Schedule K-1. Futures held inside the fund may be subject to Section 1256 rules in the US. Canadian-listed commodity ETFs have their own structures and tax treatment. See how futures are taxed in the US, and read the fund’s prospectus. This is not tax advice.

What can go wrong

Roll drag: in contango, a futures ETF can lose steadily even if spot prices are flat.
Premiums and discounts: if creations stop, the market price can move far from NAV, as USO’s 36% premium showed.
Forced changes in holdings: position limits, accountability levels, and broker limits can push a large fund away from its stated benchmark.
Daily reset decay: leveraged and inverse funds can lose value in choppy markets even if the benchmark ends flat.
Structural events: reverse splits, suspensions, and fund closures can happen at the worst moment.
Headline mismatch: “oil is up 10% this year” does not mean an oil futures ETF is.

The short version

Futures ETFs hold futures, not the physical commodity, and must roll them before expiry. In contango, each roll buys less exposure, so returns lag spot prices; in backwardation, the reverse. USO’s 2020 filings show how far a large fund can drift from its benchmark when markets break: suspended creations, a 36% premium to NAV, rapid changes in which contracts it held, a reverse split, and a 67.66% NAV decline for the year against a 20.54% fall in its benchmark. Leveraged and inverse funds add daily resetting, which makes their multi-day returns path-dependent.