The Commitments of Traders report is a snapshot taken at Tuesday's close and published on Friday afternoon. By the time a retail trader reads it, the market has had three full sessions to do something else. Every serious use of the COT report starts from accepting that lag rather than pretending it away, and most of the bad uses of sentiment data come from forgetting it.
Sentiment tools split into three families with very different data behind them. Regulatory position reports, which are audited counts of real positions with a delay. Broker-published client positioning, which is fast but is one venue's retail book. And derived measures such as options skew or volatility indices, which infer mood from prices rather than counting anyone.
The COT report and who is in each bucket
The US Commodity Futures Trading Commission publishes aggregated open interest in futures markets, broken into trader categories. For financial futures, including the currency contracts that trade on the Chicago exchange, the traders in financial futures report splits participants into dealers and intermediaries, asset managers and institutional accounts, leveraged funds, and other reportables. For physical commodities the disaggregated report uses producers and merchants, swap dealers, managed money, and other reportables.
The categories matter more than the totals. A commercial hedger holding a large short in a commodity is not bearish, it is a producer selling forward against inventory, and reading that as a directional view is the most common misuse of the report. Leveraged funds and managed money are the closest thing in the data to a speculative view, and it is their positioning that is worth tracking across time as a percentile of its own history rather than as a raw number. Our dedicated guide to reading the COT report works through the columns.
The other structural caveat is coverage. Spot foreign exchange is traded over the counter and does not appear in any position report. What the COT captures is the futures market, which is a real but partial reflection of currency positioning, and one dominated by a particular set of participants. Treat it as a window into a room rather than a view of the building.
Broker sentiment, and what it actually samples
Several brokers publish the share of their clients currently long and short an instrument, and aggregators such as the community outlook on Myfxbook pool linked accounts across brokers. The appeal is immediacy: it updates continuously and it describes people who are in the market right now.
The limits are structural. It is one venue's clients, or in the aggregator's case the subset who chose to link an account publicly. It usually counts accounts or positions rather than notional value, so a hundred micro lots weigh the same as one large position. And the population is retail, so it tells you about crowd behaviour rather than about market positioning. Where that data comes from matters too, which is part of what our piece on Myfxbook and account verification covers.
The contrarian reading is popular and only partly justified. Retail books do tend to sit heavily against strong trends, because buying dips in a downtrend is the instinct that a losing account is built on. That regularity is documented enough that some traders use extreme retail readings as a caution against joining the same side. It is a tendency and not a rule, and building a system on it is a different proposition from using it to double-check an idea.
No sentiment measure predicts price. Positioning can become more extreme for weeks before it unwinds, and trading against a crowd that is right for another month is expensive. Trading is high risk and these tools do not reduce it.
The derived measures
The third family reads mood from prices instead of counting positions:
- Risk reversals in currency options, which compare the implied volatility of calls and puts at the same distance from spot and show which side the market is paying up to protect.
- Volatility indices, which express the market's priced expectation of future movement and tend to spike when positioning is being forced out.
- Put and call ratios on equity index options, which describe hedging demand and are noisy around expiry.
- Investor and fund manager surveys, which are honest about being opinion rather than position, and which move slowly.
These update faster than any position report and carry no reporting lag, at the cost of being an inference. A spike in the price of downside protection tells you what people are paying for, which is close to but not identical to what they own. Read alongside the volatility measures you already track, they mostly answer the question of how nervous the market is rather than which way it will go.
How to use any of this without hurting yourself
Three habits separate the traders who get something out of sentiment data from the ones who get stopped out by it. The first is to plot the measure as a percentile of its own multi-year history, because a raw net position number means nothing without the range it usually occupies. The second is to check whether the extreme has begun to reverse, since a crowded position that is starting to unwind is a different situation from one still being added to. The third is to take the entry from price and structure rather than from the reading itself.
Practically, this belongs in a weekly routine and not on a chart you watch intraday. Friday afternoon after the report, or Sunday alongside the week's calendar review, is enough. Note which instruments show stretched positioning, mark those as candidates where a news surprise could produce an outsized move, and size accordingly. That is the honest use case: sentiment as a hazard map, telling you where the market is likely to move violently if something breaks.
What it will not do is tell you that a top is in. Crowded is not the same as wrong, and the number of traders who have shorted a strong trend because a positioning report looked extreme is large enough that the pattern deserves its own warning label.
"Positioning tells you where the fire exits are crowded. It does not tell you when the alarm goes off."
— Alex Onta, Executive Director, SINGUARD
Key Takeaways
- The COT report snapshots Tuesday's close and publishes on Friday, so every reading is three trading days old before you see it.
- Commercial hedgers are not expressing a view, so read the speculative categories separately rather than the net total.
- Broker long-short percentages sample one venue's retail clients by position count, not the market by notional value.
- Plot any sentiment measure as a percentile of its own history, and take entries from price rather than from the reading.
Frequently Asked Questions
What is the COT report and when is it published?
The Commitments of Traders report is published weekly by the US Commodity Futures Trading Commission. It shows aggregated open positions in futures markets broken down by trader category, and the snapshot is taken at Tuesday's close and released the following Friday afternoon US time. That gap of three trading days is the single most important thing to remember when using it.
Does broker sentiment data show what the whole market is doing?
No. A published long and short percentage reflects the clients of that one broker, weighted by the way that broker counts positions rather than notional size. It is a sample of retail activity at a single venue, which can be useful as a crowd gauge but is not a measure of market positioning. Institutional flow is invisible in it.
Can sentiment be used as a timing signal?
Poorly. Positioning can stay stretched for weeks and become more stretched before anything unwinds, so an extreme reading is context rather than an entry. Traders who use it well treat it as a filter on ideas generated elsewhere, or as a warning that a crowded position may unwind violently on news, and they take entries from price.