Every number we publish comes from official, public, primary sources — and every transformation we apply to them is described on this page. If something here is unclear, that is a bug: tell us.
Sources
- CFTC Commitments of Traders — weekly futures positioning, published Fridays for the preceding Tuesday. We use Managed Money for commodities and Leveraged Funds for financial futures.
- FINRA Reg SHO daily files — per-stock short-sale volume (paid letter).
- OCC daily volume — equity and index put/call ratios (paid letter).
- FINRA margin statistics — monthly margin debit balances (paid letter).
The core calculation
Raw contract counts mean nothing on their own — 50,000 contracts net long is crowded in one market and quiet in another. So we rank each market’s current speculative net position against its own trailing three years and report the percentile: 0 means the most short in three years, 100 the most long. Above the 90th percentile we call it crowded long; below the 10th, crowded short. We also flag any weekly change larger than two standard deviations of that market’s normal weekly change.
How to read the states
Every market gets one of five states, based on where today’s speculative net position ranks inside its own last three years (the percentile). Color is direction — green means positioned long, red means positioned short. Depth is intensity — the darker the shade, the more crowded the bet.
crowded long
Percentile 90–100. Speculators are more bullishly positioned than in roughly 90% of the last three years. New traders: this does not mean “buy” — it means the optimistic bet is already very popular, so a lot of buying power has been spent. Pros: stretched positioning is fuel — if the story disappoints, crowded longs unwind fast and exits get narrow. Watch for the percentile rolling over, not the level alone.
leaning long
Percentile 70–90. A clear bullish tilt, but with room to extend. Trends often live here for months — a lean is normal in a healthy trend and is not by itself a warning.
neutral
Percentile 30–70. Positioning is unremarkable versus its own history. Neutral means positioning gives you no edge in this market this week — whatever your view, it will have to come from somewhere else. Knowing where you have no signal is as valuable as knowing where you do.
leaning short
Percentile 10–30. The mirror of leaning long: a clear bearish tilt, not yet extreme.
crowded short
Percentile 0–10. The bearish bet is about as popular as it has been in three years. New traders: heavily shorted markets can rally violently on good news, because shorts must buy to exit — the “short squeeze”. Pros: pair the percentile with the weekly change (the z column) — a crowded short that is still growing is a different animal from one that has started covering.
The two numbers next to the state: the 1w change column shows how positioning moved since last week, and z expresses that move in standard deviations of that market’s normal weekly change — anything beyond ±2 is an unusual week worth understanding. A big z against a crowded state (say, crowded long with sharply negative z) is often the most interesting cell in the whole table: the crowd has started to leave.
What this is — and is not
Positioning is a crowding and context gauge, not a timing signal. Extremes can persist for months. COT data is as of Tuesday and published Friday — a three-day lag, minimum. We publish descriptive readings, never buy or sell recommendations, and we hold no positions in anything we cover at the time of writing. Nothing on this site is investment advice.