This page in one lineBig funds bet on whether prices will rise or fall. This page shows which way they are betting in twelve markets, and how unusual that is. The short answer: when they all piled in before, they were right about as often as wrong.
What Big Funds Are Betting On
Big funds bet on whether prices go up or down. Every week the US regulator publishes what they are holding. This is the 2026-10-06 report. The most unusually crowded bet on a rise right now is Nasdaq 100, at 100 out of 100. The most unusually crowded bet on a fall is Euro, at 2 out of 100. That leaves 3 of 12 markets unusual enough to be worth a second look: Nasdaq 100 at 100, Bitcoin at 90, Euro at 2. Now the honest part. When a bet on a rise got this crowded in the past, the price did worse than usual over the next three months in 5 of the 10 markets we have enough history for. When a bet on a fall got this crowded, the price did better than usual in 3 of 9. Those counts are out of 10 and 9 rather than 12 because a market is left out when it has been crowded on fewer than five separate occasions, which is too few to read anything into. Neither of those is more than half, so on this evidence crowding is no better than a 50-50 guess at what comes next. The crowd was right about as often as it was wrong.
Which way are the big traders leaning right now?
Think of each market as a rope. On one end are the big speculators betting the price goes up; on the other, those betting it goes down. The bar shows who is winning that tug of war, as a share of all the contracts in the market. A bar to the right means more big money is betting on a rise. The darker a bar, the more unusual that lean is for that particular market.
Drag the chart sideways to see the rest of it
Weekly report dated 2026-10-06. Bitcoin’s reading, for example, is +12%: more big money is betting on a rise than a fall.
How unusual is today’s lean for each market?
A lean of +20% might be normal for gold and extreme for the yen. So each market is compared only with itself. Here is the sum, so you can check it. Each market has about 156 weekly readings over three years. Line them up smallest to biggest and find where this week sits. If it sits 148th out of 156, that is 95, so the crowd score is 95. A score of 100 is the most crowded bet on a rise in three years, 0 the most crowded bet on a fall, and 50 is completely normal.
Drag the chart sideways to see the rest of it
Only a score of 90 or more, or 10 or less, counts as crowded on this page. Everything between those two is normal and tells you nothing.
What this does not mean. A high crowd score does not mean the price must fall. It means almost everyone who wanted to bet on a rise already has, so there are fewer buyers left. Our own record below shows that turned out badly for the crowd less than half the time, so do not read a high score as a sell signal.
How has the Bitcoin lean moved against its price?
Two lines from the same weeks. The top line is the big speculators’ lean in Bitcoin futures. The bottom line is the Bitcoin price. Watch for moments when the lean was at an extreme and what the price did in the weeks after.
Drag the chart sideways to see the rest of it
What this does not mean. Bitcoin futures on this exchange are one slice of Bitcoin trading. The large crypto exchanges outside the US are not in this report.
What is the crowd outside the US doing?
The weekly report only covers one US exchange. Most Bitcoin and Ether futures trade on big exchanges elsewhere, so here is the same question asked of them: on the largest of those exchanges, what share of trading accounts are betting on a rise? Half and half is neutral. The US big-speculator bar from the main chart is shown underneath each for comparison. Darker bars are more stretched against their own last three years.
Drag the chart sideways to see the rest of it
Offshore readings taken on the same Tuesdays as the report. Offshore records start 2020-06-02. Bitcoin: 52% of accounts are long, -2 points versus four weeks ago.
What this does not mean. Counting accounts is not the same as counting money. One large account betting on a fall can outweigh thousands of small accounts betting on a rise.
How much money is riding on it, and who has been getting hurt?
Every bet needs someone on the other side, and a bet stays open until one of them closes it. Add up all the bets nobody has closed yet and you get the size of the game. A bigger game means more people in it, which makes price swings larger. Some people bet with borrowed money. If the price moves against them far enough, the exchange closes their bet for them and they lose what they put in. The second chart shows which side that happened to over the last four weeks: bars to the right mean the people betting on a rise were the ones getting closed out.
Drag the chart sideways to see the rest of it
What this does not mean. Big liquidation numbers sound scary but they are a normal part of these markets. Read the share, not the size.
What did ordinary investors and the wider market do over the same four weeks?
Two more crowds for context. First, US spot ETF buyers: ordinary investors and advisers who buy Bitcoin and Ether through the stock market. Their net buying or selling over the last four weeks is shown in dollars. Second, the big-picture backdrop: the US dollar, the 10-year bond yield, the stock-market fear gauge and the oil price, each as its change over the last four weeks. One thing to know before reading these: the betting figures higher up this page come from a weekly government report, so they are a few days old by the time you see them, while these four are the newest readings available. Each one is dated, so you can see exactly how fresh it is.
Drag the chart sideways to see the rest of it
Each of these four comes from a different official release, published on different days, so each one shows the date it was read. Every figure here is the newest one available. US dollar index: 121.38 on 2026-10-02, against 118.07 on 2026-09-04 (the Federal Reserve's broad measure, normally 110 to 125); 10-year US bond yield: 5.22 on 2026-10-08, against 4.95 on 2026-09-10 (the interest the US government pays to borrow for ten years; above 4.5% is high by the last twenty years); Stock-market fear gauge (VIX): 15.41 on 2026-10-08, against 17.84 on 2026-09-10 (below 15 is calm, above 25 is frightened); Oil price: 96.24 on 2026-10-06, against 94.21 on 2026-09-08 (the price of one barrel of US crude oil; roughly 60 to 90 has been the usual range in recent years, so anything above that is expensive).
What this does not mean. ETF flows are reported with a one-day delay and do not include buying outside the US.
When the big traders were this crowded on a rise before, what happened next?
For every past week where the lean was crowded on a rise, look at where the price was thirteen weeks later. A past week only counts as crowded if it looked crowded at the time, judged against the three years before it, so nothing here uses hindsight. The first chart is the typical result after those crowded weeks (typical means the middle value, so one wild week cannot drag it). The second is the typical result after any week at all, for comparison. If the first is lower than the second, crowded bets on a rise tended to disappoint.
Drag the chart sideways to see the rest of it
What this does not mean. A weaker-than-usual three months is often still positive. Crowded does not mean crash.
And after a crowded bet on a fall?
The mirror image: weeks where the crowd score was 10 or less, and the price thirteen weeks later. When bets on a fall are crowded, the market can rise on little news because the sellers have already sold.
Drag the chart sideways to see the rest of it
What this does not mean. Some markets have only a handful of crowded-fall weeks in their usable history, and a run of consecutive crowded weeks is really one example. Small samples swing wildly.
All twelve markets in one table
Three groups are shown. Big speculators bet on direction. Hedgers are businesses protecting themselves, so they usually lean the opposite way. Small traders are everyone else. The change column shows how the big speculators’ lean moved over the last four weeks.
Drag the chart sideways to see the rest of it
| Market | Big funds | Businesses protecting themselves | Everyone else | Crowd score 0–100 | 4-week change, points | 3 months after a crowded rise | 3 months after a crowded fall | 3 months after any week |
|---|---|---|---|---|---|---|---|---|
| Nasdaq 100 | +20.5% | -24.1% | +3.6% | 100.0 | +13.4 points | +4.7% typical move, on 10 separate occasions, and the price rose 68 times in every 100 | +7.6% typical move, on 10 separate occasions, and the price rose 89 times in every 100 | +6.0% |
| Bitcoin | +11.9% | -14.0% | +2.2% | 90.4 | +4.6 points | -9.3% typical move, on 14 separate occasions, and the price rose 32 times in every 100 | +79.8% typical move, on 6 separate occasions, and the price rose 82 times in every 100 | +7.4% |
| Ether | +8.2% | -10.1% | +1.8% | 81.4 | -1.8 points | +18.6% typical move, on 8 separate occasions, and the price rose 70 times in every 100 | too few times to say | +1.1% |
| Copper | +22.8% | -25.6% | +2.8% | 80.8 | -8.2 points | +8.2% typical move, on 10 separate occasions, and the price rose 73 times in every 100 | +3.3% typical move, on 5 separate occasions, and the price rose 88 times in every 100 | +4.0% |
| Gold | +53.1% | -61.7% | +8.6% | 80.1 | -3.3 points | +8.3% typical move, on 8 separate occasions, and the price rose 80 times in every 100 | +2.3% typical move, on 6 separate occasions, and the price rose 53 times in every 100 | +3.9% |
| Japanese yen | +16.9% | -16.8% | -0.2% | 72.4 | +14.8 points | -0.7% typical move, on 9 separate occasions, and the price rose 42 times in every 100 | -3.5% typical move, on 10 separate occasions, and the price rose 26 times in every 100 | -1.7% |
| US dollar | +23.5% | -27.9% | +4.4% | 68.6 | -6.9 points | +4.9% typical move, on 7 separate occasions, and the price rose 95 times in every 100 | -0.1% typical move, on 10 separate occasions, and the price rose 49 times in every 100 | +0.2% |
| US 10-year bond | -15.6% | +15.1% | +0.5% | 43.6 | +0.4 points | -1.0% typical move, on 13 separate occasions, and the price rose 28 times in every 100 | -2.0% typical move, on 18 separate occasions, and the price rose 23 times in every 100 | -0.7% |
| Silver | +22.0% | -39.6% | +17.5% | 28.8 | -3.2 points | +1.5% typical move, on 13 separate occasions, and the price rose 58 times in every 100 | too few times to say | +2.9% |
| S&P 500 | -8.0% | +2.2% | +5.7% | 21.2 | -4.3 points | -4.1% typical move, on 7 separate occasions, and the price rose 23 times in every 100 | +4.1% typical move, on 9 separate occasions, and the price rose 72 times in every 100 | +4.8% |
| Oil | +5.2% | -6.7% | +1.5% | 16.7 | -1.8 points | too few times to say | +0.8% typical move, on 18 separate occasions, and the price rose 54 times in every 100 | +0.4% |
| Euro | -11.4% | +8.3% | +3.1% | 1.9 | -6.9 points | too few times to say | too few times to say | -0.0% |
What this does not mean. The report is published with a three-day delay and covers one exchange per market. It is a snapshot of who holds what, not of who is right. We start from 13 markets; 1 are left out here because they do not yet have enough weekly history to be scored against three years of their own readings.
What should you do with this?
Two kinds of reader use this page. One runs money for other people and has rules to follow. The other is deciding about their own savings. The same evidence leads to different actions.
If you run money for others
Funds, trading teams, the people who manage a company’s cash, research teams.
- A market you hold has a crowd score of 90 or moreLook up that market's own row in the table before doing anything. On this page's record a high score was followed by a weaker three months in fewer than half the markets, so it is not a reason to sell. Treat it as a reason to check how much you hold, and to stop adding.
- You are betting on a fall in a market with a crowd score of 10 or lessEveryone else is betting on a fall too. That is when a sudden rush of buying (a squeeze) is most likely, so decide in advance the price at which you will give up the bet, before the next report, not after.
- The lean moved more than 10 points in four weeksAsk what changed. A fast swing in what the big traders hold usually comes before a change in how the market reacts to news.
If it is your own money
Anyone deciding what to do with their own savings.
- You are about to buy something the big traders are already crowded intoSlow down and look at that market’s own bar in the after-crowded chart. Where crowded rises faded before, buying smaller or later cost little. Where they kept going, the crowd was right and the lesson is different.
- A coin or commodity you own is at a crowded-fall readingThis is not a reason to panic-sell. Check that market’s bar in the after-crowded-fall chart before deciding anything.
- You want a single weekly habitLook at the crowd score column once a week. Middle readings mean this report has nothing to tell you. Even at the extremes our own record says the crowd was right about as often as wrong, so at most this should change how much you hold, not by betting everything.
The conclusion
A plain reading of the weekly report on who holds large futures positions, for twelve major markets from Bitcoin to bonds.
Prices move when the last buyer or seller runs out. Knowing how crowded one side already is tells you how much room is left.
It gives one number per market, the crowd score, which says how unusual today's lean is for that market, and it shows honestly that past extremes did not reliably predict what came next.
Nasdaq 100 is the most crowded bet on a rise and Euro the most crowded bet on a fall. 3 markets are stretched. Historically, a crowded rise was followed by a weaker three months than usual in 5 of 10 markets, an even split.
Words used on this page
Every technical word above is explained again here, in plain English.
- FuturesA contract to buy or sell something at a set price on a future date. Big traders use them to bet or to protect themselves.
- Big speculatorsLarge traders, such as big investment funds, who hold futures to bet on the direction of price.
- HedgersBusinesses such as miners, farmers, refiners or exporters who use futures to protect against price moves in something they already own or need.
- Small tradersEveryone whose position is too small to be reported individually.
- Net position or leanContracts betting on a rise minus contracts betting on a fall, as a share of all contracts in that market.
- Crowd scoreHow unusual this week’s lean is for that one market. Line up its roughly 156 weekly readings from the last three years, smallest to biggest, and see where this week sits. Sitting 148th of 156 gives a score of 95. 100 is the most crowded bet on a rise in three years, 0 the most crowded on a fall, 50 completely normal.
- CrowdedA crowd score of 90 or more, meaning crowded on a rise, or 10 or less, meaning crowded on a fall. Anything in between is not crowded.
- 13 weeksAbout three months. The page checks what price did 13 weeks after each report.
- PositioningWho holds what: the mix of bets on a rise and bets on a fall in a market at a given moment.
- ExchangeThe marketplace where these futures contracts are traded. The report covers one regulated US exchange per market.
- PointsPercentage points. A lean that moves from +10% to +14% of the market has moved 4 points.
- TypicalThe middle value of a list, also called the median. One wild week cannot drag it.
- Betting on a rise or a fallTraders call these going long and going short. The page uses the plain words.
- SqueezeA sudden rush of forced buying when many traders who bet on a fall all have to get out at once, pushing the price up fast.
- StretchesRuns of consecutive crowded weeks. Ten crowded weeks in a row are one example, not ten.
- Offshore accountsTrading accounts on the large crypto exchanges outside the US. The share betting on a rise is counted by account, not by dollars.
- Open betsEvery bet needs someone on the other side, and it stays open until one of them closes it. This is all the bets in a market that nobody has closed yet, added up.
- Liquidations or forced closuresBets that the exchange closed automatically because the price moved too far against them.
- Spot ETFA fund on the stock market that holds the actual coin, so ordinary investors can buy it through a broker.
- Fear gauge (VIX)A number from the options market that rises when stock investors are paying up for protection.
- Bond yieldThe yearly interest a bond pays as a share of its price. The 10-year US yield is the cost for the US government to borrow for ten years.
- US dollar indexThe dollar’s value against a wide basket of other currencies, including smaller economies, so a rise means a stronger dollar. This is the Federal Reserve’s broad measure. It sits around 118, whereas the narrower dollar index quoted in most news stories sits around 100. Same dollar, different basket, so do not compare the two numbers directly.
- Trading daysDays the US stock market is open, which is when the ETFs trade.
- Crowd score, offshore exchangesSame idea as the crowd score above, but it ranks the share of accounts betting on a rise against that market’s own last three years.
Where this page could be wrong
- The report covers one regulated exchange per market. For crypto especially, most trading happens elsewhere.
- It is published on Fridays for the previous Tuesday, so it is always a few days old. Prices here are taken on the Tuesday; a reader could only have acted on the Friday.
- Bitcoin’s records start in 2018 and Ether’s in 2021, and every market needs two years of readings before it can be scored at all, and is then ranked against the three years up to that week, so the crypto history tests rest on fewer weeks than the older markets.
- Some markets have only a handful of crowded weeks in the usable history. A run of crowded weeks in a row counts as one occasion, and where there were fewer than five separate occasions no result is shown at all.
- Crowded weeks are rare by definition, and they come in runs. In some markets there are only a handful of separate crowded stretches, so the after-crowded results are rough.
- Positions are counted in contracts, not dollars, and contract sizes differ between markets.
- Past reactions to crowded readings are tendencies, not promises.
- The offshore reading counts accounts on one large exchange, not dollars, and its records only go back a few years.
- ETF flow data arrives with a one-day delay and covers US funds only.