Source: https://forexiapro.com/cot-report/what-is-a-cot-extreme/

# What Is a COT Extreme? Index 20 and 80 Explained, and How Often They Happen

Dylan Forexia14 November 20267 min read

What is a COT extreme? It is a reading where the large speculators are near the end of their own range. We score them from 0, the most short they have been in the last 156 weeks, to 100, the most long, and call an index of 20 or under, or 80 or over, crowded. That is our line, not the CFTC's, and a crowded reading says where the crowd stands, not when price turns. On eight markets it is not rare: it covered 46% of the weeks we counted.

[![What is a COT extreme: a crowd's position drawn against its own range, with crowded short at the bottom, an ordinary middle, and crowded long at the top](/seo/what-is-a-cot-extreme/what-is-a-cot-extreme-the-two-ends-of-the-range-diagram.webp)](/seo/what-is-a-cot-extreme/what-is-a-cot-extreme-the-two-ends-of-the-range-diagram.webp)

Diagram: an extreme is the edge of the crowd's own range, short at one end and long at the other. The real counts are below.

The report is the CFTC's weekly count of who is long and who is short in each futures market, and the large speculators are the group its legacy report calls non-commercial. They are funds, not the retail herd. [How to read the COT report](/cot-report/how-to-read-the-cot-report/) works a real row, and [what the COT index means](/cot-report/cot-index-meaning/) shows how the number is built. Here we ask what the line means and how often it is crossed.

## What is a COT extreme, in one sum?

It is the large speculators' net position, as a share of open interest, placed between the lowest and highest share of the last 156 weeks:

**Index = (this week's share − lowest share) ÷ (highest share − lowest share) × 100**

On the Australian dollar, counted Tuesday 20 October 2026, the crowd was net short 127,292 contracts, 38.2% of open interest. Over the 156 weeks the share ran from −52.6% to +30.9%, so (−38.2 + 52.6) ÷ 83.5 × 100 gives an index of 17. That is under 20, so it was a crowded-short reading, and the first since 2 December 2025.

Three things follow from the sum.

- **It compares a market with itself.** An index of 20 on the pound and one on the kiwi do not describe the same size of position.
- **It is not the direction.** The Canadian dollar's crowd at 34 was still net long. An index says how near the edge, not which side of zero.
- **It depends on the ruler.** The index moves when old weeks leave the window, with nobody trading.

## How often is a market at a COT extreme?

More often than most people expect. We counted the 156 reports to 29 September 2026 on the seven majors and gold, turned round where the pair needs it, which is 1,248 market-weeks in all. At 20 and 80, 570 of them were crowded, 46%. At the tighter lines of 10 and 90, the other pair readers use, 299 were, 24%.

| Market | Crowded weeks, 20 and 80 | Crowded weeks, 10 and 90 |
| --- | --- | --- |
| EUR/USD | 39 (28 short, 11 long) | 17 (17 short, 0 long) |
| GBP/USD | 79 (50 short, 29 long) | 41 (28 short, 13 long) |
| USD/JPY | 66 (35 short, 31 long) | 42 (27 short, 15 long) |
| USD/CHF | 57 (0 short, 57 long) | 25 (0 short, 25 long) |
| AUD/USD | 78 (44 short, 34 long) | 44 (15 short, 29 long) |
| USD/CAD | 91 (9 short, 82 long) | 51 (6 short, 45 long) |
| NZD/USD | 79 (74 short, 5 long) | 42 (38 short, 4 long) |
| Gold | 81 (1 short, 80 long) | 37 (0 short, 37 long) |
| All eight | 570 of 1,248 | 299 of 1,248 |

Read it market by market, because each has habits. At 20 and 80 the euro was crowded in 25% of its weeks and USD/CAD in 58%. Gold and USD/CHF were net long in all 156 weeks. A crowded long is close to their normal state, so we treat the read as weaker there. The kiwi leans the other way, crowded short in 74 weeks, with the large speculators net long the pair in only 23 of the 156.

So tightening the lines roughly halves the count and keeps the same shape. It changes how often you are told something. We have not tested whether it changes how right the telling is.

## How long does an extreme last?

Often for weeks, and sometimes most of a year. A run is an unbroken stretch of weeks crowded on the same side. At 20 and 80 over the same 156 reports there were 82 runs, the median was 5 weeks, 22 lasted 10 weeks or more, and the longest was 27, on the pound, crowded short from 17 February 2026. At 10 and 90 there were 56 runs, the median was 4, 10 lasted 10 weeks or more, and the longest was 19, USD/CAD crowded long from 16 April 2024.

[![What is a COT extreme: two markets' large-speculator index over 156 weeks, the euro ranging from net long to the most short in three years while gold stays net long every week](/seo/cot-report/cot-index-euro-and-gold-three-years.webp)](/seo/cot-report/cot-index-euro-and-gold-three-years.webp)

EUR/USD and gold, large speculators, 156 weekly reports to Tuesday 27 October 2026, from the CFTC's legacy futures-only figures. The top row is the net position as a share of open interest. The bottom row is the index. The euro ends at 0 and gold at 60.

The chart shows two very different markets. The euro's crowd swings from net long to the most short it has been in three years. Gold's stays net long in every week, so its index moves inside a long-only range.

## What does a COT extreme not tell you?

When, or whether. Two real cases show it.

**The pound at index 0.** The pound's crowd was at 0 in the reports counted on 29 September, 6 October and 13 October 2026, the most short it had been in three years, and the pound fell across that run before it rose. An index of 0 did not mean the pound had to rise by any date.

**Gold at the top.** The report counted on Tuesday 25 August 2026 had the crowd net long 243,334 gold contracts, index 98, with gold closing at 4656.63. On Wednesday 26 August gold's low came at 11:15 New York at 4582.78, after US inflation and growth figures at 08:30, and the day closed at 4589.97, $42 back under last week's high of 4631.94. Gold printed 4445.20 on Friday 28 August and 4282.03 at 23:30 on Tuesday 1 September, $415 under its top.

[![What is a COT extreme: gold on 26 August 2026 with the crowd long at index 93, closing $42 back under last week's high of 4631.94 at 4589.97](/seo/cot-report/gold-crowded-long-cot-report-26-august-2026.webp)](/seo/cot-report/gold-crowded-long-cot-report-26-august-2026.webp)

Gold (XAU/USD), 15-minute candles, Wednesday 26 August 2026. Gold had run $65 over last week's high by Tuesday. The low came at 11:15 New York at 4582.78 and the day closed at 4589.97. The report published on 21 August had the crowd long at index 93.

The count of 1 September had the crowd at 228,124 net long, index 93, with gold closing at 4327.43, $329 under the 25 August close. So the crowd barely moved through a large fall. It stayed near the top, and the report said nothing about when that would end. That is a reading of where they stood, and a day seen afterwards is not a setup.

## Why do traders say extremes matter, and what do we do with one?

Because a crowd at the end of its range has little room left to add, and has something to lose if price turns. That is our reading, and the report holds no stop orders. At 20 or under, or 80 or over, we read the large speculators as the side not to join, which is [Fade the Crowd](/glossary/fade-the-crowd/). It rules out joining a crowded side. It does not make a trade against it.

From [the three seats](/glossary/the-three-seats/), one reading gives three pictures. Retail sees conviction, everyone short so short must be right. The market maker, as we read it, sees orders to come, because a short has to be bought back one day, often from a stop above the highs. The observer sees which side is ruled out and nothing about when.

What we have tested: over 16 years of weekly reports on the seven majors and gold, trades taken with the crowd, joining the side the large speculators held at an extreme, lost in 12 of the 16 years. Fading it on a level hunt showed a small positive effect, which is promising and not proven. The test used 20 and 80. Our pages on [whether the COT report works](/research/) have the figures with their caveats.

## What goes wrong when people read an extreme?

Four errors come up most.

1. **Treating 80 or 20 as a signal.** It is a label of where the crowd stands. [The COT report strategy](/cot-report/cot-report-strategy/) page shows one crowded reading with two paths.
2. **Comparing across markets.** An 80 on gold, crowded long in 80 of 156 weeks, says less than an 80 on the euro.
3. **Forgetting the ruler.** On gold, seven points of a slide from 87 to 80 came from old weeks leaving the window, not from any trading.
4. **Using a stale reading as a current one.** The report is three days old on release, and [is the COT report lagging](/cot-report/is-the-cot-report-lagging/) has what that costs.

To work the index for any market yourself, use the [COT index calculator](/tools/cot-index-calculator/). Members read each week, with the stance worked out, on the [COT Report page](/cotreport), and Course 1 of ForexiaPRO, [Reading the Map](/pro), teaches the read in its module named Fade the Crowd.

## Frequently asked questions

### What is a COT extreme?

A reading where a group of traders is near the end of its own range. We score the large speculators from 0, the most short they have been in the last 156 weeks, to 100, the most long, and call an index of 20 or under, or 80 or over, crowded. The CFTC publishes no index, and the lines are our choice.

### What is a good COT extreme level, 20 and 80 or 10 and 90?

Neither is a law. Across the seven majors and gold over the 156 reports to 29 September 2026, 20 and 80 flagged 570 of 1,248 market-weeks, 46%, and 10 and 90 flagged 299, 24%. Tighter lines mark fewer and more extreme weeks. They do not make the reading more accurate, and we have not tested that.

### Does a COT extreme mean price will reverse?

No. It says where the crowd stands, not when or whether price turns. The pound's crowd was at index 0 in three reports in a row before the pound rose, and gold's crowd stayed near the top through a $329 fall in a week.

### How long does a COT extreme last?

Often longer than people expect. Over the same 156 reports the median run at 20 and 80 was 5 weeks, 22 of 82 runs lasted 10 weeks or more, and the longest was 27 weeks, on the pound, short.

### Is an extreme the same on every market?

No. Gold's crowd was net long in all 156 weeks and crowded long in 80 of them, so a high reading on gold says less than the same one on the euro, which was crowded long in 11 weeks.

## More from the blog

### [Gold M and W Pattern: Sixteen Real XAUUSD Days, Two in Full](/gold-trading/gold-m-and-w-reversals/)

The gold M and W pattern on real days: sixteen marked XAUUSD days from May to September 2026 with the second leg and the reversal, and two worked in full.

### [How to Trade Gold on FOMC Day: A Real Marked-Up Day, 17 June 2026](/gold-trading/gold-on-fomc-day/)

How to trade gold on FOMC day, read on a real marked-up XAUUSD chart: the 14:00 decision, who was trapped, and three more Fed days set beside it.

### [Forex Weekly Outlook: How We Map the Week on Sunday](/weekly-cycle/forex-weekly-outlook/)

Our forex weekly outlook starts with last week's high and low, the COT stance and the calendar, not a forecast. One real Sunday and what it could not say.