Commitment of Traders open-interest extremes as regime filters

What the report publishes

The Commitment of Traders report publishes open interest for a contract and assigns traders to commercials, noncommercials, nonreportables, and spreaders once an exchange reporting threshold is crossed.

Traders below the reporting limit remain nonreportable. Traders above it must declare commercial hedge status or large-speculator status, while spreaders hold longs and shorts in different delivery months.

Open interest analysis uses those published totals and category shares. The report is a weekly record of who holds the open interest, not a verdict on which group is better informed.

Measure who is adding or reducing exposure

Charts of commercial net longs can sit near historical extremes at major price bottoms while noncommercials sit net short. That snapshot does not by itself prove commercials are always the better-informed group.

Tracking week-to-week changes shows noncommercials often reach those short extremes by selling throughout a decline, while commercials reach long extremes by scaling into hedges on the way down.

A weekly estimate of each category's outcome can be built from the change in that category's net long position, the period's price change, and contract size. Summing those weekly estimates across the studied natural-gas history is consistent with a liquidity-for-certainty market-role model rather than a simple smart-money ranking.

Weight extremes by share of open interest

A participation-weighted extreme index can be assembled from each group's position relative to its own history, scaled by that group's share of open interest excluding spreaders.

When that oscillator approaches its high readings, category positions are on average near extremes and an existing trend may be closer to reversal or consolidation. The same construction is illustrated on corn as well as energy.

Pair the index with moving-average direction

The extreme index is nondirectional like an average directional movement reading, so it is meant to be paired with a moving-average system for buy-versus-sell context.

Lookbacks and extreme thresholds should be recalibrated as participant behavior changes.