CFTC Commitments of Traders
Where the big speculative funds sit in corn, soybeans, wheat, cattle, hogs and milk this week, and how crowded that is. Not a price forecast.
Data: CFTC Commitments of Traders, released Fridays for positions as of the Tuesday before.
Who the funds are, and how old these numbers are
These are the traders the CFTC calls managed money: registered commodity trading advisors, commodity pool operators and unregistered funds that trade futures for clients. They trade paper and never own a bushel. Crowded positioning tells you how much buying or selling could be forced if the market turns, not which way it turns. The test on the full record is below.
Positions are counted on Tuesday and published Friday afternoon (the Monday after, when a federal holiday falls Wednesday to Friday that week), so they are at least three days old when they land and older every day after. The figures below carry their own age. Does positioning predict price? We test our own rule, in public, one call per crop each week.
Fund positioning by commodity
Where every market stands
Each market placed by how crowded fund positioning is against its own record, and by what price has done since. Position is measured as a share of all open contracts, so an old reading and a new one mean the same thing in a market that has since changed size. The dashed lines are the 20th and 80th percentiles; a dot is coloured only when the market is crowded (top tenth of its record while net long, bottom tenth while net short). Bottom right is the corner to watch: funds on the high side while price has already turned down.
What moved this week
Ranked by the size of the week's change against each market's own open interest, so a move in corn contracts and a move in Chicago wheat contracts, which are markets of very different size, can be compared. The last column is that change scored against this market's own history of weekly changes.
Scatter plot of fund positioning percentile against thirteen-week price change for every tracked market. The same figures are in the table below.
Show the board as a table
Does this predict prices?
The test on the full record: does fund positioning line up with later price moves?
Measured on the full CFTC record. Correlation runs from −1 to +1; zero means no relationship.
Managed money net position: 52 weeks
One market at a time
Pick a market. Every figure is drawn from the full CFTC record. Forward returns are measured from the first close after the report was published, which is not always the Monday: a federal holiday late in the report week moves it.
Loading.
Why our numbers are lower than the ones you'll see elsewhere
Four mistakes that make a positioning study look better than the trade it describes, and how this page avoids them.
Four mistakes make a positioning study look better than the trade it describes. They all push the same way, and they are all easy to make by accident.
- Measuring from a price that had not printed yet. Positions are held as of Tuesday. The report usually comes out Friday at 3:30 pm Eastern, after the grains have settled for the week; a federal holiday from Wednesday to Friday of that week moves it to the Monday. A study that measures from the Tuesday close gives itself three days of the move it claims to have predicted. Every return here starts at the first close after the report existed.
- Reading a contract roll as a price move. A continuous front-month chart steps by the whole spread when the front month changes. Corn's July to September roll often spans old crop and new crop, and the step can be a few cents one year and far more in a tight one. Those steps are removed here before anything is measured. Lean hogs are the exception: the calendar that finds their roll is not yet correct, so no price read is shown for hogs until it is.
- Counting overlapping weeks as separate evidence. Thirteen-week returns sampled every week share twelve weeks of the same price path. Eighty matches can be six actual spells. Each read below prints both numbers.
- Searching until something is significant. Many tests run every week, and at the usual threshold one in twenty looks real by chance alone. The whole batch is counted and corrected before anything is called a finding.
What does this mean for farmers?
What the report is, how to use it, what managed money means and where the position sits.
What is the COT Report?
The CFTC publishes Commitments of Traders usually on Friday at 3:30 pm Eastern (on the Monday after when a federal holiday falls on the Wednesday, Thursday or Friday of the report week; a Monday holiday does not move it), counting positions held as of the previous Tuesday. It splits the market into four reportable categories: managed money, producers and merchants, swap dealers and other reportables, plus a small-trader remainder. This page leads with managed money, which is the speculative money, and breaks out all five further down. Every contract has two sides, so the categories net to zero: when funds buy, somebody sold.
How to use it
As a measure of who is exposed. It does not call direction: on the whole record (the test is above), positioning and price move together in the week they are measured; the relationship with the next four to thirteen weeks is close to zero. What a crowded position does tell you is how much buying or selling could be forced if the market turns, which is a reason to protect unpriced bushels rather than a reason to predict a direction.
What "managed money" means
CTAs, commodity pools and hedge funds. They do not own grain, they trade paper, and their buying and selling moves the futures leg. It does not follow that it moves your basis. Basis is your local cash price minus futures, and it is set by what is happening at your elevator: bushels arriving, space, freight and demand. Fund flow and basis frequently move in opposite directions.
Where the position sits
The bar on each card covers the last year. The sections above rank the same position against the whole record we hold, measured as a share of all open contracts, which is the comparison that survives a market doubling in size. When the card and the section disagree they are answering different questions: one about this year, one about the whole record.