DetectorKilled

The liquidation-cascade detector that predicts the wrong way

−0.56 bps

the average move after a cascade, prices bounce back, so the bet was the wrong way

bps = basis points = a hundredth of a percent (100 bps = 1%).

In plain English

Our cascade detector fired perfectly, but prices bounce back instead of continuing, so it pointed us to bet in exactly the wrong direction.

What it needed

prices keep moving after a cascade (worth ~18 bps)

→

What it got

−0.56 bps, prices bounce back instead

Why it's dead

Liquidation cascades mean-revert, they don't continue. The detector fires perfectly, pointed in exactly the wrong direction.

The detail

n=15,992 silent cascades. Mean return is negative at every hold (−0.35/−0.56/−0.86/−2.63 bps at 1m/5m/15m/60m), with density gradient at Spearman ρ=−1.0 (denser cascade → more reversion). Even at zero cost the best in-sample magnitude (+0.56 bps) is an order of magnitude under the ~18 bps hurdle.

Kill date: 2026-03-26
Sample: n=15,992
Method: Pre-registered
Verdict: right mechanism, wrong sign

What it would have done to your money

You put in$10,000

You run it for100 trades

You would have

$9,944

−$56

lost (1%)

You started with$10,000

This strategy left you$9,944

Instead of not trading$10,000

Locked to this strategy's real measured result of -0.0056% per trade, compounded. You choose the amount and the time. We don't choose the return, the strategy already did.

Pre-registered before the data. Judged on a criterion locked in advance. Published whatever the result.

See all kills →