See what traders are
actually doing with their money.
Not opinions or forecasts. We measure real open positions and published filings, then show you who is buying and who is selling.
Four separate signals, each on its own clock: ordinary traders at the brokers we track, the tone of the news, big investment funds reported to the US regulator, and what central banks are saying. Where those disagree is usually the interesting part.
Reading the live positioning data…
Live positioning is loading. If this line stays, the data service is unreachable and the page will recover on its own. Everything below still applies.
Most interesting first
Sorted by how far apart the ordinary traders and the big funds are. Read the top row and you have the story.
All four signals, on every pair
Here you can see what ordinary traders are doing on all 11 pairs, and all four signals on EURUSD, USDJPY and XAUUSD. The FXC Terminal opens the two slow signals everywhere else.
- Funds and Policy on every pair and currency we track
- Each signal's own history, not just today's number
- What each individual broker book says
- Alerts when the crowd flips or the funds reach an extreme
Free readings stay free. Nothing on this page moves behind the wall.
Who are these two groups?
The whole page is built on measuring these two, and they behave very differently.
People trading from a phone or laptop at a retail broker. We read the brokers that publish their client positioning and take the middle value.
updates every 30 minutesHedge funds and professional money managers. Their positions are reported to the US regulator once a week, so this number is always a few days old.
updates weeklyWhat you are looking at
Sentiment on this page is not a forecast and not a score. It is a measurement of four separate groups of people: what they are holding, and what they are saying out loud. We keep them apart, because they often disagree, and the disagreement is the useful part.
| The group | What we read from them | How often it changes | Where you get it |
|---|---|---|---|
| Ordinary tradersCrowd | The share of open positions on each side, at 9 retail brokers | Every 30 minutes | Every pair, free |
| Big fundsFunds | The net position of leveraged funds, as filed with the US regulator | Once a week, always 3 days behind | 3 instruments |
| Central banksPolicy | How firm or how soft their own published language reads | When they speak, sometimes not for weeks | 3 instruments |
| The pressNews | The tone of coverage, weighted by how important each story is | Continuously, and quiet when nothing is written | Every pair, free |
How each number is made
The crowd number is the one most people come for, so here it is in full.
Each broker publishes what share of its clients are buying.
Not the average. One odd broker cannot drag the number around.
Published exactly as read. We never flip it and never guess a missing one.
Nine brokers report anywhere from 41% to 78% buying. We publish the one in the middle, 61%. We also show how far apart the nine are, because that is worth knowing on its own: nine brokers agreeing inside a point is a very different picture from the 37 point spread you see here, even though both would publish a single figure.
We take the net position from the regulator's weekly filing and place it in its own trailing 52 week range. That way "high" means high for this market, rather than a contract count you would have to have memorised.
Every statement, set of minutes and speech is scored on one fixed scale for how firm or how soft the language reads. The scale never changes, so two banks and two years are comparable.
Every story is scored for tone and weighted by how important it is, over a rolling 48 hours. A major story moves the reading, a routine one barely touches it.
Why we never average them into one number
One number would be simpler to read, and the temptation is real. Here is what it would cost.
Two groups on one side, two on the other. Four units, four clocks, shown next to each other and never folded together.
And here is what it threw away
- That two of the four groups are firmly on the opposite side
- That not one of the four is actually neutral, though the average reads that way
- Which parts moved this morning and which have not moved since the last meeting
They are not the same kind of number. The crowd reading is a share of open positions. The funds reading is a place inside its own 52 week range. Policy is a score for how firm a sentence reads. News is a weighted tone. To average them you first have to pretend all four are the same unit, and they are not.
They also move at completely different speeds. Retail positions turn over in hours. A fund's filing is three days old before you ever see it. A central bank's stance can hold for a quarter. Blending a number that changed this morning with one that has not changed since spring produces something that is neither current nor slow, and there is no honest label to put on it.
And it destroys the only part worth having. When the groups split, as they have above, the split is the finding. One averaged figure hides it by construction, every single time, and it hides it hardest exactly when the disagreement is widest. That is why the board on this page is sorted by how far apart the groups are rather than by the size of any one number.
How to use it
Sentiment tells you where people are positioned. What you do about that is your call, and we deliberately do not make it for you.
65% short does not mean the price is about to rise. It means the book is one sided, and one sided books behave differently: the stops are stacked on one side, and there are fewer people left to keep pushing. Whether you read that as a reason to fade the crowd is your decision. We publish the number as reported and never flip it.
52% is noise, and most pairs sit near it most of the time. 78% is a crowded book. The same is true of the funds reading, which is why we show it against its own 52 week range instead of as a raw number: it is the distance from normal that carries the information.
The most useful view here is where two groups sit on opposite sides. Ordinary traders heavily long while the funds are short is a genuinely different market from both leaning the same way, even when the headline percentage is identical. It is the first thing to check and the reason there are four readings at all.
Going into a central bank meeting, a funds position filed three days ago may already describe a world that has gone, while the crowd number is half an hour old. Both are correct measurements. One is simply more current, so every reading carries the time it was measured and you can see for yourself.
What it does not tell you
Worth being blunt about, because the four timestamps invite a reading they do not deserve.
- When anything will happen None of the four readings has a horizon. A crowded book can stay crowded for a week or unwind inside an hour, and nothing on this page tells you which one you are in.
- The timestamps are publishing schedules, not forecast windows A weekly stamp on the funds reading means the US regulator publishes weekly. It does not mean the reading is about next week. A 30 minute stamp on the crowd does not make it a 30 minute forecast. The clock belongs to the source, never to a prediction, because there is no prediction here.
- Where the price may go next That is a different question and it is not answered on this page. It gets answered separately, as one monthly reading with a measured record attached rather than an opinion: how stretched a market is, and what followed before.
- What to trade Nothing here is a trade recommendation, and it will not become one.
Positioning is the wrong input at that speed. What the spread is doing against its normal for this hour, how much the market is actually moving against its own typical range, and how many minutes remain until the next release all matter far more, and all three are measured separately from this.
This page describes what market participants currently hold and say. It is measured state, not a forecast, and nothing here is a trade recommendation.