How to read COT data

The raw COT file gives you a dozen columns per trader group: long, short, spreading, changes, percentages, trader counts. Most of it is noise for a discretionary trader. Three numbers carry almost all of the meaning, and they are best read in this order.

1. Net, not gross

A group's net position is its long contracts minus its short contracts. Positive means the group is positioned for a rise, negative for a fall.

Gross numbers mislead. A fund category can hold a large long book and an equally large short book — the size looks dramatic and says nothing. The net is what remains after both sides cancel, and it is the number that moves when the group changes its mind.

Start with one group per market: leveraged funds in financial futures, managed money in commodities. These are the discretionary and systematic funds — the participants whose job is to be right about direction. The other groups are informative, but they are the ones taking the other side of that trade.

2. The change beats the level

A net of −74,000 contracts tells you where the group stands. The week-on-week change tells you what it is doing, which is usually the more actionable half.

A position that grows in the same direction for several weeks shows conviction building. One that is being cut week after week shows a trade going out of fashion — often while the price still looks fine. And a large position that stops growing is a different animal from one that is still growing: the fuel is there, but nobody is adding.

3. The 0–100 index puts it in scale

Here is the problem with raw contract counts: 100,000 contracts is an enormous position in one market and a rounding error in another. Even within one market, what counted as “a lot” two years ago may be normal today.

So positioning is normalised against its own history. Take the net position over the last 52 weeks, find the lowest and the highest, and express today's value as a position between them on a scale of 0 to 100. That is the COT index:

1. Near 100 — the group is more long than at any point in the past year.
2. Near 0 — more short than at any point in the past year.
3. Near 50 — ordinary. Whatever is driving the market this week, it is not positioning.

The index makes two things possible at once: comparing gold with the euro without doing arithmetic, and spotting when a market's own crowd has reached the edge of its usual range.

A worked example

These are the numbers from the most recent report, for gold (the full page is here). Managed money — the fund category in commodity markets — held 149,721 long contracts and 12,950 short as of Tue, Sep 1, 2026.

1. Net first. 149,721 minus 12,950 is a net of +136,771 contracts. That single number replaces two, and it is the one worth tracking.
2. Then the change. A week earlier the net was +144,747, so funds trimmed the position by 7,976 contracts.
3. Then the range. Over the past 52 weeks this net moved between +89,752 and +163,519. Today's +136,771 sits at 64 out of 100 in that range — in the middle.

Three numbers, in that order, and you know more about gold positioning than the price chart can tell you: how funds are placed, whether they are still moving, and whether the current level is unusual for them.

What extremes do and do not mean

When positioning reaches one end of its range, the practical statement is narrow and worth getting exactly right: most of the people who wanted this trade already have it. That makes the market vulnerable to a squeeze, because new buying has to come from somewhere and the existing holders are the ones who can be forced out.

What it does not give you is timing. Extremes can persist for months and grow more extreme along the way. Positioning describes a state of the market, not a moment — and treating it as an entry signal is the single most common way to lose money with this data.

Four mistakes worth avoiding

1. Reading hedgers as forecasters. Producers and swap dealers are usually on the other side of the funds by construction. Their net is a mirror of the speculative position, not a competing opinion.
2. Comparing contracts across markets. Only the 0–100 scale is comparable; the contract counts are not.
3. Forgetting the three-day lag. Positions are Tuesday's, published Friday. In a week with a shock, the report describes the world before it.
4. Using one week in isolation. A single change is noise more often than not. The trend over four to eight weeks is where the story is.

Where to go from here

Every market page on this site shows the same three numbers in the same order — net position, weekly change, and where that sits in the 52-week range — with the reasoning written out for the current week. That is what WAVG does with the raw file: read it the same way every week, for every market, so the comparison is honest.