F-0010 · first recorded 2026-08-21 · updated 2026-08-24
In CASCADE-1, forced-liquidation episodes did not move price more than volatility-matched random minutes in the same hour
MEASURED
| Observation | 15m mean +40.07 bps in the forced direction against a matched control of 44.52 bps; no horizon clears both benchmarks |
|---|---|
| Sample | 228 independent episodes >=$250k across 28 symbols (n=106 with price data), Binance market-wide, 2026-08-18 to 2026-08-20 |
| Method | CASCADE-1, contract frozen sha256 7dee22eed9cdaecb before any outcome; episodes not events; permutation null plus same-symbol same-hour control |
| What is not established | K1 met at 228 episodes, so this is not an underpowered null; three days, one venue, and only two of four declared benchmark tiers were implemented |
| Who else publishes this | no liquidation product publishes a volatility-matched control |
| Evidence | research/cascade-1-result.md, market/cascade1.py |
At 15 minutes the raw effect is large and would make a compelling screenshot: +40 bps, 60% hit rate, clearing a permutation null. Take any other minute from the same hour in the same symbol and you get more continuation.
In this three-day Binance sample, liquidations occurred while markets were already moving, and the test found no incremental continuation beyond the matched volatility control. That supports reading liquidation as a symptom in this sample; it does not establish that all liquidations on all venues and regimes are incapable of affecting price.
What this closes. The cascade forecast product. The defensibility metric (F-0001) and the cascade model both work and both attach to a phenomenon that did not add measurable continuation in CASCADE-1.
What it does not touch. F-0002 — the book thins to 0.846 during large moves — is a fact about microstructure and stands independently.