COT Report Strategy: Why the Crowd's Lean Is Not a Trade

A COT report strategy, on our read, starts as a no: when the large speculators are crowded, do not join them. That leaves a lean, a direction you are allowed to look in. A lean is not a trade. The report gives no time, no price and no way of being wrong, and on most weeks nothing comes of it.

This is the page for what you can do with the Commitment of Traders report once you have read it. If you are new to the number itself, start with the COT index and what 20 and 80 mean.

What are the usual COT report strategies, and what does each run into?

Most guides offer five uses, and each one runs into the same missing thing: when. This table puts them against the reports we read in 2026.

Each row is a reading, and the guides mostly stop at "combine it with price action". That sentence is where the real work starts.

What does a crowded extreme tell you?

It tells you three things, and we read the third as an inference.

• Which side not to join. At 20 or under, or 80 or over, that is the crowd's side. Our word for this stance is Fade the Crowd.
• How much is already done. A crowd near the end of its range has done most of its selling, or its buying. Few are left to push price further its way.
• Where its stops probably rest. We read each short as a stop above the highs and each long as a stop under the lows. The report holds no stop orders. This is a reading of the structure.

So it gives you a lean: under a crowded short, a direction you may look up in.

What does a crowded extreme not tell you?

It does not tell you the four things a decision needs.

• When. It has no clock. A crowded side can sit for weeks while price runs its way.
• Which path. The next move may run against the crowd or with it, and the Friday report looks the same either way.
• At what price you would be wrong. The report names no level. A lean cannot be stopped out, which is why it feels so comfortable.
• Whether anything will happen at all. Most weeks nothing like the days below comes.

Two real reports show the gap. Both were published before 2 October 2026, so ours were built afterwards, by the same rules, from the CFTC's public figures.

What did a crowded short look like? EUR/USD, July 2026

The report counted on Tuesday 21 July and published on Friday 24 July had large speculators net short 41,338 euro contracts, 5.2% of open interest, at an index of 18: crowded short. From that Friday, the report said the same thing all the next week. Price did not.

On Monday 27 July the euro fell to 1.13668, 3 pips above the low of the week before, 1.13637, and stopped. A reader had a crowded short, a falling euro and nothing to go on. On Tuesday 28 July it went to 1.13531 in the 06:00 candle, 11 pips under, with a German central bank report due that minute that our calendar marks as medium impact. To a seller that was a breakdown. We read it as the stop hunt, which most traders call a liquidity sweep, of the stops under that low. Tuesday ended at 1.13859, 22 pips back above. We read the move back as the Close Back.

The next days ran the other way: 1.14709 on Wednesday, with a Federal Reserve decision that afternoon, and 1.15369 on Thursday. The crowd had not helped anyone see that coming. The count taken on that same Tuesday shows it had added to its shorts, net short 72,447, index 8.

Set the two days side by side. On Monday and Tuesday morning the report, the lean and the direction of price were all the same, and only the chart differed. The glossary page follows the same report over the following weeks. A reversal seen afterwards is not a setup.

What did a crowded long with no such day look like? Gold, September 2026

It looked like being right with nothing to act on. The report counted on Tuesday 1 September and published on Friday 4 September had large speculators net long 228,124 gold contracts, 54.9% of open interest, at an index of 93: crowded long. On gold we count $1 as 1 pip.

Last week's high was 4510.64, from Thursday 3 September. In the twelve trading days after the report, gold's highest point was 4442.70 on 8 September, $68 short of it. There was no push over that high for anyone to chase. Gold fell anyway. On Monday 14 September it reached 4253.44 in the 09:00 candle, $257 under the high, and went through the low of early September, 4282.03. Our calendar lists no high- or medium-impact US release that day. It ended at 4298.56.

So the side-not-to-join part was fine. A buyer had a poor fortnight. Anyone who sold because the crowd was long took a guess, and a guess that came good is still a guess. The same reading two weeks earlier, on 18 August at 93, was followed by gold running higher for days first.

The euro's week and gold's fortnight started from the same kind of Friday. The report could not separate them.

Why is a lean not a trade?

A trade is a decision with a time, a price and a point at which it is wrong, and a weekly table of Tuesday's positions contains none of the three. The lean narrows what you look at. Everything after that is built on the chart, and how we build it stays in the course.

Each of the three seats met 28 July differently. Retail saw the push through 1.13637 as the move starting. The market maker, as we read it, saw the crowd's last orders filling. The observer could not say at 06:00 which it was. The euro was back above that level by the candle opening at 07:15, and the report had still not said what came next.

We tested it, too. Run over many years, trades taken with the crowd mostly lost, and fading it on a level hunt showed a small effect that is promising and not proven. Separately, trading against the stop hunt as a mechanical rule did not make money in our own test. So the read explains a position. It does not make the next trade pay.

What mistakes do traders make with a COT report strategy?

All four come from asking the report for more than it holds.

1. Buying the euro on Monday 27 July because the crowd was short. It closed that day at 1.13675 and was lower still the next morning. A seller of gold on the same reasoning in September saw gold fall, and the report could not tell those two apart.
2. Buying gold because it read 93 on 1 September. That number meant the crowd was already long, and in the twelve trading days after the report gold never got back over 4510.64.
3. Treating the fall as proof. Gold's fall did not show that the read works. One fortnight, picked afterwards, shows one fortnight.
4. Trading the report on Monday. The positions are six days old by then, and the crowd may have moved. The next count often says so, too late.

Keep a record if you want to know more than this page can tell you. Each Friday note which side is crowded on one market, then write what price did over the next fortnight. Most entries will say nothing happened, and that is the useful part.

Course 1 of ForexiaPRO, Reading the Map, teaches this read step by step in lesson 1.3.3, The Lean and the Trade.