# Aave Post Reports 301× Underpricing; Pairwise Results Were Significant at Less Than 0.001

Published 2026-11-09 · Crypto.world Team · Crypto.world

> A forum post reports 57 days with four simultaneous liquidation triggers, compared with 0.19 expected under an independent model. It also reports that all 36 pairwise asset combinations were significant at less than 0.001.

## Four assets hit on the same day 57 times, versus 0.19 expected

On 2026-11-09, oraclethaacat (@oraclethaacat) posted that [Aave](https://governance.aave.com/t/per-asset-liquidation-thresholds-underpredict-co-liquidation-by-301x/26131) V3’s per-asset liquidation thresholds underpredict co-liquidation by 301×. The test counted 57 days when four assets hit liquidation triggers on the same day, compared with 0.19 expected. Aave V3 computes liquidation risk per asset.

## The test covered nine major collateral assets

The analysis covers nine major collateral assets and 2,123 trading days from September 2020 to October 2026, using Aave’s published parameters and public daily price data from Yahoo Finance. For each asset, the test marked days when a position at maximum LTV would cross the liquidation threshold. It then shuffled each asset’s data column independently, using 2,000 permutations to break cross-asset dependence while preserving each asset’s marginal rate.

## Five-asset triggers and six-or-more-asset triggers

For five assets, the test found 31 real days with simultaneous triggers against 0.004 expected. For six or more assets, it found 21 real days and none in 2,000 shuffles. All 36 pairwise asset combinations were significant at less than 0.001. The post identifies May 19, 2021, May 11, 2022, and Nov. 8, 2022, as the worst days, citing all nine assets on the first date, Terra/Luna on the second and FTX on the third.

## The post proposes a joint probability model

The post calls for replacing per-asset thresholds with a joint probability model that accounts for tail correlation across collateral assets. It says portfolio-level liquidation thresholds should tighten as correlated assets are added, rather than average. The post says the mathematics was developed after 2008 because the same independence assumption failed in CDO pricing. It characterizes the problem as structural inadequacy in the thresholds, not a code bug: Aave’s code correctly implements the specified thresholds.

## Sources

1. Original source [Aave Governance Forum](https://governance.aave.com/t/per-asset-liquidation-thresholds-underpredict-co-liquidation-by-301x/26131) governance.aave.com

## Companies in this story

- [Aave Labs](https://crypto.world/companies/aave-labs)

## Related stories

- [Aave proposal would narrow V3 Core to established collateral](https://crypto.world/news/aave-proposal-narrow-v3-core-collateral)
- [LlamaRisk recommends Aave V3 cap changes on Mantle, Monad, BNB and Core](https://crypto.world/news/aave-risk-stewards-propose-v3-cap-cuts-mantle-monad)
- [Threshold Pauses tBTC Operations as Aave Freezes Ethereum Reserve](https://crypto.world/news/threshold-pauses-tbtc-aave-freezes-ethereum-reserve)
- [Aave Freezes 292 Reserves After Exploiter Borrows $44.6M](https://crypto.world/news/aave-freezes-292-reserves-binance-exploiter-borrows-44-6m)
- [Aave Labs reports 6,096 issues completed in its first 180-day update](https://crypto.world/news/aave-labs-first-180-day-dao-service-report)

## Explore

[More articles](https://crypto.world/news)

---

Canonical: https://crypto.world/news/aave-forum-post-301x-co-liquidation-risk-gap
