“When the big speculators are all on one side, the move is nearly over. Look for the reversal.”
Verdict: not supportedEvery week the CFTC publishes how big speculators (hedge funds and other large traders) are positioned in currency futures. When that positioning reaches an extreme, it's often described as a "crowded trade" that's due to reverse. Satdish publishes a COT breakdown every week, so we wanted to know whether extremes are actually worth acting on.
| What we measure | Speculators' net position in each currency's futures |
| What counts as extreme | In the top 10% or bottom 10% of its own last 3 years |
| When we act | Close of the first trading day after the report is published |
| The trade | Crowded long: sell the currency. Crowded short: buy it. Hold 8 weeks |
| Costs | 0.02% every time the position changes (about two pips) |
The timing matters. The report shows positions as of Tuesday but isn't published until Friday afternoon, so a backtest that trades on Tuesday is using information nobody had yet. Ours waits until after it's public.
What each currency did against the dollar over the next 8 weeks. If the claim were right, the "after crowded long" column would be clearly negative and the "after crowded short" column clearly positive.
| Currency | Weeks crowded long | Next 8 weeks | Weeks crowded short | Next 8 weeks | Any week |
|---|---|---|---|---|---|
| Euro (EUR) | 185 | +0.80% | 178 | −1.40% | −0.03% |
| Pound (GBP) | 212 | −0.70% | 170 | −0.95% | −0.15% |
| Yen (JPY) | 186 | −0.30% | 236 | −0.29% | −0.17% |
| Aussie dollar (AUD) | 203 | −1.10% | 141 | −0.85% | −0.62% |
| Canadian dollar (CAD) | 139 | −0.08% | 181 | +0.51% | +0.01% |
| Swiss franc (CHF) | 126 | +1.20% | 189 | +0.98% | +0.34% |
| Kiwi dollar (NZD) | 100 | −0.94% | 188 | +0.71% | −0.06% |
There's no consistent reversal. Some currencies drifted the "right" way after extremes, others the wrong way, and most moves are small. The euro is the clearest counter-example: after speculators were crowded long, it kept rising (+0.80% on average), and after they were crowded short it kept falling (−1.40%). On the euro, the crowd was usually right.
The contrarian trade on all seven currencies at once, equally weighted, after costs.
2008 shows the danger. In the financial crisis, speculators piled into positions and the moves kept going. Fading them meant standing in front of the biggest trends of the decade.
| Currency | Yearly return | Sharpe | Worst drawdown | Sharpe 2004–14 | Sharpe 2015–now |
|---|---|---|---|---|---|
| Euro (EUR) | −3.35% | -0.46 | -64% | -0.66 | -0.19 |
| Pound (GBP) | −0.55% | -0.07 | -40% | -0.18 | +0.04 |
| Yen (JPY) | −0.93% | -0.11 | -43% | -0.20 | +0.01 |
| Aussie dollar (AUD) | −0.52% | -0.06 | -40% | -0.13 | +0.04 |
| Canadian dollar (CAD) | +1.37% | +0.26 | -24% | +0.09 | +0.47 |
| Swiss franc (CHF) | +0.55% | +0.07 | -27% | -0.16 | +0.29 |
| Kiwi dollar (NZD) | +1.84% | +0.22 | -37% | -0.01 | +0.49 |
Fading the euro lost 3.3% a year, the one result here strong enough that it's unlikely to be chance. The Canadian and New Zealand dollars did a little better than break-even. Nothing is strong enough to build a strategy on.
Nine variations: looser and stricter definitions of "extreme", and shorter and longer holds. The highlighted row is the main rule. This is a check, not a hunt for the best setting.
| Extreme means | Hold | Sharpe, whole period | 2004–14 | 2015–now |
|---|---|---|---|---|
| Top 20% / bottom 20% | 4 weeks | +0.01 | -0.29 | +0.38 |
| Top 20% / bottom 20% | 8 weeks | -0.05 | -0.33 | +0.28 |
| Top 20% / bottom 20% | 13 weeks | -0.04 | -0.23 | +0.16 |
| Top 10% / bottom 10% | 4 weeks | +0.08 | -0.30 | +0.57 |
| Top 10% / bottom 10% | 8 weeks | -0.02 | -0.34 | +0.37 |
| Top 10% / bottom 10% | 13 weeks | +0.03 | -0.26 | +0.36 |
| Top 5% / bottom 5% | 4 weeks | +0.02 | -0.23 | +0.36 |
| Top 5% / bottom 5% | 8 weeks | +0.04 | -0.11 | +0.24 |
| Top 5% / bottom 5% | 13 weeks | -0.06 | -0.21 | +0.12 |
Two things stand out. Every version is close to zero over the whole period. And every version lost money in the first half and made a little in the second. That's a consistent pattern and genuinely interesting, but it's modest, it only covers about eleven years, and one decade working after another failing is exactly what luck can look like. It's something to keep watching, not something to trade.
For interest: each currency's current rank against its own last three years (2026-09-22 report). Given the results above, treat this as context, not a signal.
| Euro (EUR) | 11th | Normal range |
| Pound (GBP) | 4th | Crowded short |
| Yen (JPY) | 80th | Normal range |
| Aussie dollar (AUD) | 56th | Normal range |
| Canadian dollar (CAD) | 74th | Normal range |
| Swiss franc (CHF) | 61th | Normal range |
| Kiwi dollar (NZD) | 62th | Normal range |