A friend asked me why I still bother with a government report that shows up three days stale, and the honest answer is that the Commitments of Traders report is the closest thing traditional futures markets have to a block explorer. Every week the CFTC publishes who is positioned where, in size, across currencies, metals, energy, grains, and equity index futures. You do not get individual accounts. You get every large trader bucketed by type, and the buckets turn out to be the useful part. I have seen people pay for sentiment dashboards that are mostly this free file with a coat of paint on it.
What the report actually contains
The mechanics shape everything else, so start there. Any trader holding positions above the CFTC reporting thresholds gets counted, the agency takes a snapshot as of Tuesday's close, and the aggregated data comes out Friday afternoon US time. So you are always reading positioning that is roughly three days old. In a quiet market the lag barely matters. In a week where something broke on Wednesday, the report describes a world that no longer exists, and you have to hold that in your head every time you open it.
There are two report families. The legacy format splits traders into three groups: commercials, who use futures to hedge a business exposure, non-commercials, the large speculators, and non-reportables, everyone too small to hit the threshold, which works as a rough proxy for retail. The disaggregated format, which covers physical commodities, splits things more usefully into producers and merchants, swap dealers, managed money, and other reportables. Financial futures get a parallel version called Traders in Financial Futures, with dealers, asset managers, leveraged funds, and others. If you only learn one, learn the disaggregated one. Pulling swap dealers out of the commercial bucket fixed the legacy report's biggest distortion, since a dealer hedging a bank's commodity index exposure is economically carrying speculative flow even though the old format filed it under hedging.
One more choice, futures only versus futures plus options combined. The combined version converts options into futures equivalents, and I default to it because large traders express plenty of views through options, though most weeks the two versions tell the same story.
How to read each group
Commercials are the people who touch the physical thing. The producer locking in prices for next year's output, the airline hedging fuel, the food company buying wheat forward. They are usually net short in commodity markets because production hedging tends to dominate, and they typically get shorter as price rises, since higher prices are better prices to hedge at. Their selling is a business decision rather than a price call, so reading routine commercial selling as bearish misses what they are doing. Where they get interesting is at extremes. When the people who know the physical market best are carrying the smallest short they have held in years, they are quietly telling you prices look cheap to the industry itself.
Managed money is the opposite animal. These are hedge funds and CTAs, a big chunk of them trend followers, so their net position mostly tells you what price has already done. They tend to be right in the middle of a trend and maximally exposed at the turn. The useful signal is crowding. When managed money net length hits a multi-year extreme, the marginal buyer has mostly already bought, and the market gets fragile to anything that forces those positions out. Setups like that have historically marked durable turns in gold, in crude, and in currency futures, with the caveat that marking a turn and timing a turn are different claims. Extremes can keep building for months before anything breaks.
Small specs, the non-reportables, are mostly noise week to week. I only pay attention when they lean hard the same way as managed money at the same time, because that tells you the whole crowd is leaning together, retail included, and those moments are rarer.
A weekly routine that takes about twenty minutes
- Pull the data after the Friday release. The CFTC publishes it free, and full historical files are available for backfilling.
- For each market you follow, compute the net position for commercials and managed money, then normalize it. The standard trick is a COT index, the current net position expressed as a percentile of its range over roughly the past three years, so every market reads on the same zero to one hundred scale.
- Divide net positioning by total open interest as well. A record net long in contract terms can be unremarkable once the market itself has grown.
- Flag anything above roughly the ninetieth percentile or below the tenth. Those go on a watchlist, and only a watchlist.
- Act only with confirmation. An extreme plus a price reversal, a failed breakout, or a real catalyst is a trade idea. An extreme on its own is a condition.
- Track week-over-week changes too. A crowded position being unwound quickly often says more about the next month than the level itself does.
The discipline that matters most is refusing to fade an extreme just because it exists. Positioning shows you where the fuel is stacked, and something else has to light it, usually price action that forces the crowded side to start covering.
Where the report breaks
The lag is the obvious problem. Positions are as of Tuesday, you read them over the weekend, and by Monday they are almost a week old. Any violent midweek move means the crowd you are studying may already be gone. Releases also occasionally stack up around holidays and government shutdowns and then arrive in batches, which quietly wrecks naive backtests that assume a clean weekly cadence.
The bigger structural limitation is coverage. The report only sees futures and options on US exchanges. For currencies that is a real issue, since the overwhelming majority of FX volume trades over the counter in spot and forwards, so currency futures positioning is a small sample you are treating as a proxy for global sentiment. It has worked reasonably well historically, especially at extremes in the yen and the euro, but hold it loosely. For metals and energy the futures market sits much closer to the center of price discovery, and the report carries more weight there.
Classification is fuzzy too. A firm registers as commercial or not, and everything it does gets stamped with that label, so a commercial that decides to punt on direction still reads as hedging. And extremes are stubborn. I have watched managed money sit at a record position for a full quarter while price ground further against the fade. A process that shorts every ninetieth percentile reading with no trigger and no stop will eventually hand you a loss big enough to make you quit the report entirely, which would be a shame, because used with patience it is one of the better free sentiment tools around.
My own setup is unglamorous, a spreadsheet of COT indexes for about a dozen markets refreshed on weekends, sitting next to the whale wallet and insider disclosure feeds I watch in Blockcircle, since the underlying question is the same everywhere. Who is positioned how, and what happens if they are forced to change their mind. Start with two or three markets you already trade, build the percentile view, and let the extremes come to you. Most weeks the report tells you nothing actionable, and accepting that is most of the skill.