Trading Liquidation Anomalies with Trend Filters and AI Review
Summary
This document outlines a crypto perpetual futures strategy that treats unusually large liquidation flows as directional signals. It builds a per-token baseline from historical liquidation amounts, compares recent activity with that baseline using a Z-score, and requires liquidations to be concentrated on one side before signaling. The proposed interpretation is trend-following: a wave of long liquidations points to a short, while short liquidations point to a long.
Signals are enriched with recent candlestick behavior and news, then passed to an AI model for a combined entry decision. The described workflow also covers recurring data collection, position selection, trailing and fallback stops, and a dashboard for monitoring positions and signals. The article gives implementation excerpts and an illustrative dashboard, but does not present a systematic performance evaluation. Results may depend on baseline construction, thresholds, exchange data quality, execution conditions, and the AI judgment; the article’s claim that liquidation cascades tend to continue is an assumption to validate, not established evidence.
Key ideas
- Liquidation amounts can be aggregated by token and compared with a historical baseline to identify unusual activity.
- A Z-score anomaly is only treated as directional when liquidations are sufficiently concentrated on one side.
- The strategy follows liquidation pressure, shorting after long liquidations and buying after short liquidations.
- Candlestick alignment and news review are added as filters before an AI-assisted entry decision.
- Trailing and fallback stops are proposed, but the article does not establish profitability through rigorous testing.
Tags
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.