AAVE Risk Assessment Tool

Check your liquidation risk on AAVE v2

Enter Ethereum Address with AAVE v2 Borrowing Activity

Example:
0x8f55AF387183855A432B61301C703A4584BC8B8D
0x4813f5074cF9DfE179ee052431CDf421F01377D2
Find more addresses here and click on any "From" address to find borrowers.
About the Risk Assessment Model
How It Works

This tool uses machine learning to predict the likelihood of liquidation for AAVE v2 users. The system analyzes user borrowing patterns, collateral assets, health factors, and historical activity to calculate a risk score on a scale of 300-850 (similar to a credit score).

The model is trained on historical data from the AAVE protocol, including past liquidation events. Using XGBoost (a gradient boosting algorithm), the system identifies patterns that lead to liquidation and assigns risk levels accordingly.

Key Features Used for Prediction

The model analyzes over 30 features, with the most important ones being:

debt_eth: Total debt in ETH value
deposit_volume_usd: Total deposits in USD
risk_heuristic: Composite risk score
health_factor: AAVE's liquidation threshold indicator
account_age_days: Account longevity
stablecoin_collateral_pct: % of collateral in stablecoins
withdrawal_to_deposit_ratio: Withdrawal vs deposit behavior
borrow_exceeds_repay_volume: Debt accumulation pattern
Model Performance

ROC AUC: 0.9918

PR AUC: 0.9658

Algorithm: XGBoost

Class Balancing: SMOTE

Feature Importance

The chart below shows which features have the greatest impact on predicting liquidation risk:

Feature Importance Chart
Model Accuracy

The following charts demonstrate the model's predictive performance:

Precision-Recall Curve

Precision-Recall Curve

ROC Curve

ROC Curve