This is the question the whole COT industry monetizes, so you should distrust anyone whose answer is a confident yes. We backtested every extreme in our 44-market universe since 2006, 1,206 of them, and published the full record: the backtest, the base rates per market, and the strategies that failed.
What the record shows
- Most extremes do not turn inside a 13-week window. Per-market turn rates are published on every file: some markets sit near 1 in 3, others lower.
- The ones that turn make you wait: median waits run weeks, with shakeouts against the fade on the way. The wait and the worst sit-through are printed on every live file.
- Mechanical rules died in testing. Fading every extreme, waiting for confirmation triggers, following the crowd: all negative after corrections in our walk-forward tests. We published those failures, because a tool that hides its failed tests is selling you the untested ones.
So what survives?
Measurement survives. An extreme is a risk state you should know about, not an instruction. What made the difference in past turns was the surrounding conditions, which is why every live extreme carries its fingerprint: how much today looks like the cases that actually turned. Measured, never forecast. The methodology holds the full test record. See one market’s full file worked end-to-end, free: the GOLD file · or what is crowded this week.