WTI Went Negative. Brent Didn't.

Daily spot prices, US dollars per barrel, not adjusted for inflation. Data: oil-prices, from the U.S. Energy Information Administration (EIA). Data story #3 — written from an outline.

On 20 April 2020 the WTI spot price was -$36.98 a barrel**. Brent, the other main crude benchmark, was **$17.36 that day. The WTI figure is the only negative value in 25,415 daily, weekly, monthly and annual Brent and WTI prices going back to 1986.

What these numbers are

Both series are EIA's daily spot prices, "free on board": Brent for North Sea crude, WTI for crude delivered at Cushing, Oklahoma, a landlocked pipeline and storage hub. They are two different crudes at two different places, not one commodity priced twice.

They are also spot prices, not futures. The widely reported **-$37.63** is the settlement price of the May 2020 WTI futures contract on NYMEX that day, according to the [CFTC](https://www.cftc.gov/PressRoom/PressReleases/8315-20). The spot figure here, -$36.98, is related but not the same number.

What the data says

• WTI was negative for one day. It fell below zero on Monday 20 April and was back above it on Tuesday.
• Brent fell hard too. On 21 April it was $9.12, two cents above its lowest daily price on record, $9.10 in December 1998. On that day WTI was back up close to it. Both benchmarks fell. Only one went below zero, and only once.
• Averaging removes it. EIA's weekly WTI price is the average of that week's trading days. For the week ending Friday 24 April, the average includes the -$36.98 and is still positive: **$3.32**.

So anyone using EIA's weekly, monthly or annual WTI series would not see a negative oil price at all. It only shows up in the daily figures.

Why, according to EIA

This dataset contains prices, not reasons. It cannot show why WTI went negative, or why Brent did not. EIA's own account, published a week later, points to the May futures contract, which expired on 21 April (per the CFTC), and to storage at Cushing. The hub was 76% full on 17 April, and some of the remaining space may already have been leased or committed. Some holders of the contract, unable to take delivery, sold at negative prices — in EIA's words, "in effect paying a counterparty to allow them to exit their positions". The Cushing spot price fell below zero on 20 April, the day before the contract expired.

How this was made

Both series come from oil-prices in this repository: EIA's own spreadsheets, rebuilt into tidy CSVs by build.ts. The one-negative-value count came from a statistics table produced by that dataset's enrich.ts. That table exists only in this repository; it is not part of any published dataset. The charts are built by oil-prices-make-charts.mjs, which reads the same CSVs; re-running it reproduces them exactly. The community datasets/oil-prices package wrangles the same EIA source independently. The two are compared in the structure benchmark.

All prices are in US dollars as EIA published them, not adjusted for inflation.

Friction notes

For the story and enrich skills:

• The consolidated table found the story. Scanning one min column across all eight resources turned up the single negative cell. Without that table it would have taken a file-by-file search.
• Outline review earned its keep. The first outline explained why WTI went negative, which a price-only dataset cannot show. It also left out Brent's $9.12 the next day, and so made the contrast look stronger than the data supports. The independent reviewer caught both before any prose was written.
• "Every number on a chart" has limits. The dataset-wide count, Brent's 1998 low and the attributed EIA and CFTC figures cannot sit on a chart of spring 2020. The skill should say which kinds of numbers are exempt.
• Author's voice pass: outstanding. This prose is an AI draft rendered from the approved outline; the "sounds like me" pass has not been done.