How to read the COT report

What the COT report is

COT means Commitments of Traders. Once a week the US Commodity Futures Trading Commission publishes how many futures contracts the large reporting firms hold, and which kind of firm holds them. It is a census of positions that already exist. It is not a forecast, and it is not a broker’s opinion of the spot price. Forex has no COT row of its own, because the spot market between banks is not a CFTC futures pit. The forex reading uses the currency future that stands in for the currency.

Euro futures stand in for the euro, so a net long there is a long in EUR/USD. Yen futures stand in for the yen, and this one flips the pair: a net long in yen futures is long the yen, which is short USD/JPY. Sterling futures map to GBP/USD, Australian dollar futures to AUD/USD. Canadian dollar futures and Swiss franc futures flip again, because those pairs are quoted as dollars per foreign currency in the future and as foreign currency per dollar in the spot pair: a net long in Canadian dollars is short USD/CAD, and a net long in francs is short USD/CHF. Dollar-index futures describe the basket, not one pair. The same census exists for gold futures, if gold is the market next to the currency book.

When it comes out, and which row to read

The positions are those of Tuesday. The CFTC publishes them on Friday at 3:30 p.m. New York time. In Paris that is often 21:30, and in London 20:30, but the US and Europe do not change their clocks on the same dates, so for a few weeks each year the release is one hour earlier. Read that Friday’s time on the calendar instead of a memorized local hour. If Friday is a US holiday the release moves, usually to the following Monday. By the time you read it, three sessions have already traded. A violent Wednesday can make Friday’s table look late. You still read it, and you say so: this is the book as of Tuesday, not the book as of this morning.

The table that matters for a directional forex read is leveraged funds. Those are the speculative books, mostly hedge funds and commodity trading advisors. The older “non-commercial” line of the legacy report is the cousin of that row, not the same row: since the disaggregated report, swap dealers sit apart. Do not add the two tables together. Commercials, in the producer and merchant group, are mostly hedgers. They are often net the other way from the funds, because they took the other side. That opposition is normal. It is not a second signal. The non-reportable line is the small traders who fall under the reporting threshold. Open interest, at the edge of the table, is the total of contracts still open. It tells you whether the market is adding positions or leaving them.

The three numbers, and the history around them

Long contracts minus short contracts is the net. A positive net is a net long. A negative net is a net short. The level alone is a poor sentence. A currency can sit net long for a year because the funds like the trend, and zero is not a magic neutral. What you compare is this week’s net with last week’s net, and this week’s net with its own range over the last one to three years. The weekly change says whether they are adding or cutting. The range says whether the trade is ordinary or crowded. A net near the top of that range is a crowded long. A net near the bottom is a crowded short. A net in the middle of the range, still growing, is a trend that still has room.

Open interest changes the verb. If the net long rises and open interest rises, new longs are being opened. If the net long falls and open interest falls, longs are being closed, which is not the same thing as fresh shorts. If the net long falls and open interest rises, new shorts are coming in. Write that sentence before you look at the candle. A one-week twitch inside an ordinary range is noise. The report is weekly, so a daily chart will always look busier than the table. You are not late to a scalp. You are reading the book the large futures traders were holding on Tuesday.

A weekly reading, in order

Do this once, after the Friday release, for the currencies you actually trade. The point of the order is to stop the eye jumping to the most exciting cell.

1. Name the future that matches the pair, and note if the pair is flipped. Euro futures for EUR/USD. Yen futures for USD/JPY, with the sign reversed.
2. Read only the leveraged-funds row: long, short, and the net. Ignore the other groups until this row is written down.
3. Subtract last week’s net from this week’s net. That difference is the only change that counts.
4. Place the net in its range over the last few years. Say crowded long, crowded short, or middle.
5. Read open interest against the week before, and say whether positions are opening or closing.
6. Write one sentence that a stranger could repeat: who is positioned, whether they added, and whether the trade is crowded. Only then look at the price.

Example: euro futures, a crowded long

This is an illustration, not a live print. Leveraged funds are long 280,000 euro futures and short 80,000. The net is plus 200,000. Last week the net was plus 170,000, so they added 30,000 contracts. The high of the last three years is about plus 210,000. The net is near that high, and it is still rising. Open interest rose with it, so these are new longs, not a reshuffle of old ones. EUR/USD can keep rising while they add. The report does not say to sell. It says the long is crowded: disappointing euro news meets more forced sellers than the chart shows, because the funds are already in.

Change one fact and the sentence changes. Same net of plus 200,000, but open interest fell and the net fell from plus 230,000. They are leaving a long, not building one. The crowded reading is fading. That is a different week from the week they added. A third case: the net is plus 40,000, close to the middle of the range, up from plus 10,000. Nothing is crowded. The report is describing a trend that is still being built. Treating that middle reading as a reversal is how a weekly table gets misused.

Example: yen futures, and the flipped pair

Another illustration. Leveraged funds are net short 140,000 yen futures. A month ago the net short was 90,000. The short sits near a multi-year extreme, and open interest rose as the short grew, so new yen shorts were opened. In the spot market that is the same trade as long USD/JPY. The chart of the pair can look strong while this is happening. The report’s sentence is: the short yen is crowded. If those funds start buying yen back, USD/JPY can fall even though the trend line is intact. You do not sell the pair because the net is large. You know who will have to buy if the story breaks.

The mistake on this pair is to read “net short” as “bearish the dollar pair” without flipping the sign. A net short in yen futures is bearish the yen and bullish USD/JPY. The same flip applies to USD/CAD and USD/CHF. EUR/USD, GBP/USD, and AUD/USD do not flip. Write the pair next to the future before you write the adjective.

What the report will not do

It will not time a Tuesday afternoon. The positions are already three days old on release, and they say nothing about the next five minutes. It will not replace the economic calendar: a payrolls Friday can overwhelm a crowded long before the next COT exists. It will not tell you the size to trade. A crowded net is a description of fuel and of risk, not an instruction. Two reporting conventions also get mixed up in blog posts. If one site shows non-commercial and another shows leveraged funds, the nets will not match, and neither site is “wrong”. Pick one row and keep it, so the history you compare is the history of that row.

Where to read it on Novex

The COT page shows that leveraged-funds net, the weekly change, and the history, with the contract mix and a search by symbol. Direction Overview places the same read next to retail positioning and the AI bias. The useful comparison is simple. If the funds and the small traders are on the same side at an extreme, the crowd is wider than the COT row alone. If they disagree, you are looking at two different books, and the sentence you wrote on Friday has to say which one you mean.

Common questions