Crude Oil Inventories

[Rodent + 1,424]

API expected to report another draw in crude oil inventories today, as US production moves back up to 9.750 million bpd. 

[Addy + 14]

A surprise build instead, both crude oil and gasoline. 

[Missy + 43]

If we can trust the API anyway. How accurate is this stuff? Seems like analysts and EIA and API never agree, sometimes to the tune of millions of barrels. I understand that it can't be 100% accurate, but how in the world can it be millions off in a single week? I guess I don't understand the intricacies of how this is all calculated.

[Marina Schwarz + 1,576]

They're doing better in the last few months. Before that, they couldn't agree on rise/fall two weeks out of every three. One wonders how they calculate these inventories, indeed.

[LAOIL + 33]

So relief for oil prices as EIA reports draw after all...why the discrepancy between API and EIA data?

[Rodent + 1,424]

Well, EIA has stricter rules for companies reporting inventories and production. There are strict penalities for misreporting data to the EIA, intentional or otherwise. The API doesn't have penalties for misreporting. However, traders have been using API data for much longer, and it's available about 18 hours earlier which is a big deal. I believe the API has about 50 more years of reporting inventories.

Also, EIA includes production data, so the data it provides is more comprehensive.

That said, I believe API and EIA are usually in lock step--I think about 80% of the time. But a couple weeks ago they were WAYYYY off, so it seems like they are not in synch at the moment. But I think that's a rarity rather than the rule.

[Addy + 14]

EIA's integrity of its weekly crude oil inventory data has been called into question in recent weeks, with the EIA defending its position by clarifying that the weekly figures are estimates based on... estimates. I read the article and was still flabbergasted at how it can be accurate, or considered accurate. But API, which is real numbers, and EIA are usually pretty close and EIA monthly figures, which I believe are real data, seems to be close. This is why the weekly numbers never add up to the monthly figures: mystery solved!