API vs EIA crude oil inventory: what's the difference

Every Tuesday afternoon the American Petroleum Institute puts out its Weekly Statistical Bulletin. Every Wednesday morning the EIA follows with its Weekly Petroleum Status Report. Both claim to tell you the same thing: how much crude and product sat in US tanks that week. They frequently don't agree, and the size of that disagreement moves WTI before the ink is dry on either one.

If you've only been trading a year or two, the first few times API and EIA print opposite signs on the headline crude number feels like a data error somewhere. It isn't. The two surveys are built differently, and the gap between them is structural, not noise.

Two different surveys, one week apart

API is a trade association. Its bulletin comes from voluntary submissions by member companies, compiled by API staff and released without the same disclosure standards a federal statistical agency operates under. There's no public methodology audit, no revision history you can pull apart line by line, and participation is the association's own member roster, not a government-mandated census.

EIA runs a different operation entirely. The Weekly Petroleum Status Report draws on Form EIA-814 and related surveys that refiners, terminal operators, and pipeline companies are legally required to file. The agency has statutory authority to collect this data, publishes its sampling and estimation methodology, and issues revisions when year-end benchmarking against the full annual survey shows the weekly estimates ran off.

That's the first source of divergence right there: a voluntary trade survey released a day early versus a mandatory federal survey with a paper trail behind it.

Why API and EIA numbers diverge

A few mechanics explain most of the weekly gap desks argue about on Tuesday night.

Sample coverage. API's respondent base doesn't cover every operator EIA mandates reporting from. Smaller independent refiners and some terminal operators that file with EIA simply aren't in API's voluntary pool. Miss a few million barrels of coverage and the headline number moves on sample composition alone, not on real stock changes.

Timing cutoffs. Both surveys try to capture the same Friday-to-Friday week, but the exact hour data gets locked in differs. A large draw or build late in the survey window can land in one report and not the other depending on when each organization's cutoff falls.

Estimation versus reported figures. EIA's weekly number is itself partly modeled. The agency doesn't get complete data from every filer every week, so it fills gaps with estimation routines and later true-ups the figure against monthly survey data that comes in with a lag. API's number carries less of that statistical smoothing, for better or worse, which means it can swing harder on a single large respondent's reported change.

Category definitions. What counts as "commercial crude" versus crude in transit, in SPR, or held at Cushing versus the broader PADD breakdown isn't bucketed identically in both releases. A barrel that shows up as a crude stock change in one report can land under a different line item in the other.

None of this means one number is right and the other wrong. They're measuring overlapping but not identical things, on different legal footing, a day apart.

What the gap actually tells a desk

The useful read is whether API and EIA agree on direction, even when they disagree on magnitude. When both show a build or both show a draw, the market usually treats Wednesday's EIA print as confirmation and trades through it quickly. When API shows a draw and EIA shows a build, that's the setup that produces the ugly five-minute whipsaw on release morning, because neither number on its own tells you which survey caught the real move.

That's also exactly the window where a position reads on something other than either agency's survey earns its keep, a stocks estimate built from what's sitting in the tanks rather than from what refiners and terminals chose to report this week. Oil Inventory Index measures floating-roof tank shadows across covered facilities from daily satellite passes and turns that into a daily CSV series, so you're not waiting on either Tuesday or Wednesday to find out which survey was closer.

If Tuesday-to-Wednesday whipsaw has cost you a position more than once, it's worth seeing what a daily, independent stocks read looks like before the next release.