How do you calculate a z-score?

Take today's value, subtract the historical average, and divide by the standard deviation of that history. A z-score of +2 means today's value sits two standard deviations above its own average.

What is a normal z-score range for COT positioning?

Most weekly readings fall between -1 and +1. Beyond ±2 the reading is genuinely unusual, historically only a few percent of weeks, and that is the zone analysts call stretched or extreme.

Why use a z-score instead of the raw number?

Raw positioning numbers cannot be compared across markets or eras: contract counts differ per market and grow over time. A z-score rescales everything against its own history, so 'extreme' means the same thing in the yen as in gold.

Is a high z-score a sell signal?

No. It says the position is rare relative to history, which raises reversal risk and reward, but rare can stay rare for months. Use it as a condition that sharpens other signals, not as a trigger on its own.

Glossary

Z-Score

What is a z-score? A measure of how far today's value sits from its historical average, in standard deviations. In positioning analysis it flags when a trade is stretched.

A z-score answers one question: how unusual is this number, compared to its own history? It rescales any value into standard deviations from its average. In positioning analysis it turns a raw net position, meaningless on its own, into a comparable measure of how stretched the crowd is, in any market, on one scale.

How it works

Three ingredients: today's value, the historical average, and the standard deviation (a measure of how much the series normally wobbles). Subtract the average from today's value and divide by the standard deviation. The result reads like this: 0 means perfectly ordinary; +1 means above average but unremarkable; +2 means higher than roughly 95% of history; +3 is genuinely rare territory.

The power is in the rescaling. Speculators net long 150,000 contracts in the euro and net long 30,000 in the kiwi cannot be compared directly; the markets are different sizes. Their z-scores can. That is why positioning dashboards, including the COT pages on this site with their 10-year z-score, lean on it: one number, same meaning everywhere.

A worked example

Suppose speculator net positioning in the pound has averaged +20,000 contracts over the past ten years, with a standard deviation of 35,000. Today the net position prints +115,000. The z-score is (115,000 - 20,000) / 35,000 ≈ +2.7. Translation: the crowd is more bullish on the pound than it has been in roughly 99% of the last decade. Nothing about tomorrow is decided by that number, but the asymmetry has changed: nearly everyone who could buy this story already has, and if the story cracks, the exit is narrow. A trader holding the same bullish view now knows to size it smaller or protect it better than the chart alone would suggest.

Common mistakes

The classic error is using the z-score as a timer: shorting a +2.5 reading because "it must revert". Stretched positioning can stretch further, and strong trends routinely spend months in extreme territory. The second error is ignoring the lookback window: a z-score against two years of history says something very different from one against ten, because a short window can make a normal reading look extreme. Know the window before you trust the number.

See it in the data

Every market on the COT positioning pages shows its current 10-year z-score next to the net position and flip percentile, so the "how unusual is this" question is answered before you ask it.

Related terms

Net Positioning CoT Open Interest

Put these concepts into practice.

See how fundamental data shapes currency bias with real-time economic indicators and sentiment analysis.