Track 18Closed
DeFi liq-map (Aave) — lending liquidations vs a CEX spike
Aave v3 LiquidationCall events predict a CEX spike — we read contract state and event topics, not raw network load
Key metric
reverse causation: the liquidation LAGS the crash (pre −110 bps ≈ the whole −116 drop); realizable fade +4.7 bps = noise
- Why an edge should exist / hypothesis
- On-chain lending gives what HL lacked: full coverage and an explicit LiquidationCall flag in the event topics. Hypothesis: an on-chain liquidation is a forced sell that overshoots and reverts on the CEX price.
- Construction
- eth_getLogs on LiquidationCall (Aave v3), matched to CEX minutes, fade the excursion, segmented thin vs thick.
- Test data
- Keyless ETH RPC (tenderly / 0xrpc / mevblocker) + CEX minutes.
- Results
- The liquidation LAGS the crash: pre-window −110 bps ≈ the whole −116 drop. Look-ahead reversion +147, but realizable fade +4.7 = noise. The thin segment is negative (−66.9 bps, WR 33% — permanent informed dumps). Thick +12 is just fading a sharp ETH drop, i.e. the existing spike edge.
- Validation
- The explicit LiquidationCall flag removed the DATA wall (full coverage) — but the economic wall is the same as HL liq-map.
- Root cause
- A liquidation is a consequence of the crash, not a cause: by the time the event fires the move has already happened (reverse causation). The thin segment is informed dumps, not fadeable.
- Verdict
- Dead ($0 cheap check). The same economic wall as HL liq-map.
- Source
scripts/research/defi_liq_fetch.py · defi_liq_cex.py · PROGRESS 2026-09-08
Want the same check on your strategy?
Validation →