How to interpret COT: net open interest, spread, and publication lag

An increase in net long positions by funds does not necessarily mean they have bought more long contracts. Similarly, an unusually large net long position does not predict when the market will peak. The Commodity Futures Trading Commission's (CFTC) Commitment of Traders (COT) report is a snapshot of the open interest in a specific futures market, grouped by trader category. Its value lies in describing market participation and concentration, provided that this data is not mistaken for real-time trading volume.

Job postings released on Tuesdays are typically posted on Fridays.

COT reports typically describe the positions as of Tuesday's close and are released on Friday at 3:30 p.m. Eastern Time. The "position date" refers to the time when the position was observed, while the "release time" refers to the time when the average reader becomes aware of it. There is usually a gap of several trading days between the two; delays due to holidays or other reasons should be checked against the actual release schedule.

Typically, reports are released at 7:30 PM Eastern Daylight Time (UTC), which is 8:30 PM Standard Time (UTC). Therefore, reports received over the weekend may miss significant developments that occurred between Wednesday and Friday. Thursday's policy surprise cannot be simply analyzed as a fund reaction to Tuesday's holdings.

Let's assume Tuesday's snapshot shows a net long position of 40,000 contracts. Prices fell on Wednesday, rebounded on Thursday, and the report is released on Friday. At this point, only Tuesday's positions are confirmed. The next report measures the change between the two Tuesdays; positions opened and closed during this period may not be reflected in the difference.

Backtesting must respect this distinction. Directly correlating the report's release date (Tuesday) with Tuesday's prices and trading based on that data introduces forward-looking bias. Both the observation date and the actual available time should be considered, and the strategy's returns should only be measured from the point where action is possible after the report's release. Even Friday's settlement price is not applicable if it was determined before the report's release.

First select the report, then select the trader category.

Traditional reporting categorizes reporting traders into two main groups: commercial and non-commercial. Physical commodity reporting further distinguishes between producers, traders, processors, and users; swap dealers; fund managers; and other reporting entities. Financial futures reporting differentiates traders by type: dealers or intermediaries, asset management companies or institutional dealers, leveraged funds, and other reporting entities.

A category primarily describes the trader's business, not the motive behind each transaction. A producer can hold positions for several purposes. A swap dealer's short futures can offset client business rather than express a bearish forecast. Asset managers may have allocation mandates, while leveraged funds may trade relative value. None of those labels establishes that every position is a directional bet.

Keep the exchange, contract-market identifier, report family and options treatment fixed when comparing weeks. Similar asset names can cover different contract sizes or venues. A sudden category change also merits checking classification or coverage changes: moving a trader between categories can alter a column without an equivalent new trade taking place.

A 10,000-contract increase in net longs can hide falling longs

Net position = long positions − short positions. That subtraction compresses two pieces of information into one. Suppose one category had 120,000 longs and 80,000 shorts last week, giving a net long position of 40,000. This week longs fall to 110,000 and shorts to 60,000. Net longs rise to 50,000.

The 10,000 increase came because shorts declined by 20,000 while longs declined by 10,000. “Net length increased, driven mainly by a contraction in shorts” fits the evidence. “Funds bought 10,000 new long contracts” does not: the reported long column actually fell.

The last two rows have identical net exposure, but their long-plus-short columns fall by 30,000 or rise by 30,000 from the initial 200,000. Participation looks very different. These sums exclude positions separately classified as spreading, so they should not be labelled the category's complete gross holdings.

Market size matters too. Net longs of 40,000 against total open interest of one million equal 4%. If net longs stay at 40,000 while open interest grows to 1.2 million, the share falls to approximately 3.33%. An unchanged contract count can represent a smaller relative exposure. A record absolute position and unusual crowding are different questions.

Contracts are not dollars of fund flows. Notional exposure also depends on price and the contract multiplier, while margin and committed capital are separate quantities. Every outstanding futures contract has a long and a short side. Across the entire market, aggregate longs equal aggregate shorts; one category's net long is offset elsewhere. The market cannot acquire an extra unmatched “net bought contract.”

Spreading and options conversion explain apparent discrepancies

For applicable categories, offsetting positions are reported separately as spreading. Imagine one trader holds 500 longs and 300 shorts in different months of the same commodity. The matched 300 can appear as spreading, leaving 200 outright longs. A net long figure of 200 alone does not reveal the size of the original two-sided position.

When reconciling open interest, add spreading to the long side and separately to the short side, once each. Omitting it can make the column totals look incomplete. Producer categories are presented differently, so not every category must have identical three-column treatment. A spreading label also does not establish that a position is riskless or perfectly market-neutral.

Futures-only and futures-and-options-combined reports measure different things. Combined data convert options into futures-equivalent positions using delta. For example, 100 long calls with delta 0.4 represent roughly 40 long futures equivalents. If their quantity stays unchanged but delta rises to 0.6, the equivalent exposure becomes 60. The extra 20 can come from sensitivity changing, without the investor buying 20 futures.

Use one treatment consistently across weeks. Conversion and rounding can also create very small differences in combined totals. Do not splice a futures-only history into a combined history, or add the two reports together: much of the underlying exposure overlaps.

Define the historical window before calling a position extreme

A commonly constructed range indicator is (current net position − window minimum) ÷ (window maximum − window minimum) × 100. If a 52-observation window has a minimum of −60,000, a maximum of 40,000 and a current value of 20,000, the reading is (20,000 + 60,000) ÷ 100,000 × 100 = 80.

That is 80% of the distance from the minimum to the maximum. It is not the 80th percentile and not an 80% probability of rising prices. A percentile requires ranking observations separately. If the maximum equals the minimum, the denominator is zero and the formula has no normal output. Changing the window or including one exceptional historical reading can alter the interpretation substantially.

Large net longs can persist through a strong trend for weeks. Selling simply because the position looks crowded may expose a trader to further gains before any liquidation occurs. Deep net shorts can also expand as demand deteriorates. An extreme is more useful for asking who might need to exit if the trend changes than for choosing the date of that change.

Use positioning as context and prices as current evidence

With net long positions near historical highs, close attention needs to be paid to whether new positive news can still push prices higher, whether recent lows will be broken, and whether subsequent reports show a contraction in long positions. Concentration, weak price response, and signs of liquidation have collectively supported the unwinding of crowded long positions. If higher highs and higher lows continue to emerge, an increase in net long positions alone is not enough to determine that a top has been reached.

If net short positions are large and prices fail to reach new lows after negative news is released, followed by a rebound to structural levels, it may indicate that short covering supported the rebound. However, by the time the weekly report is released, the rebound may have already occurred. Current support and resistance levels and the validity of any breakouts need to be examined ; a break below the starting point of the rebound would weaken signs of structural improvement.

The COT report does not show all global spot FX, physical commodity trading, or OTC derivatives exposure. Visible futures positions may only be part of a larger portfolio. Compared to intraday trading, which requires real-time price, liquidity, and event information, the COT report is better suited for studying trading participation over several weeks or months.

A useful COT reference will clearly indicate the market, category, holding date, publication time, whether it deals only with futures or both, and which data column caused the net change. Once these details are clear, the holding claim can be tested with observable market evidence.