# whether self-liquidation profitability thresholds differ systematically across lending protocols creating exploitable attack economics The self-liquidation attack documented in [[self-liquidation via flash loan is profitable when borrowed funds trigger and profit from the borrowers own liquidation before repaying the flash loan]] has a clear profitability condition: `liquidation_bonus > flash_loan_fee + borrow_fee + price_impact_cost`. Whether this condition is met depends on protocol-specific parameters that vary across lending systems. The open question: do the parameter configurations of major lending protocols (Aave V2, Aave V3, Compound V2, Compound V3, Euler, MakerDAO, Morpho, etc.) create systematically different profitability thresholds? If some protocols are consistently profitable to self-liquidate and others are not, this would constitute actionable attack economics intelligence, identifying which protocols are persistently exploitable versus which are structurally hardened. The variables requiring empirical measurement: liquidation bonus percentages (often 5-15%), borrow fee structures (variable APR vs. fixed fee), flash loan costs (Aave V3 is 0.05%), oracle deviation thresholds (how far price must move before triggering liquidation), and cool-off periods between deposit and liquidation eligibility. If systematic differences exist, they could reveal: design patterns that consistently prevent self-liquidation profitability (e.g., Aave V3's liquidation bonus caps), design patterns that consistently enable it, and whether protocol upgrades over time have moved protocols across the profitability threshold. <!-- Research direction: empirical analysis of self-liquidation economics across major lending protocols. Would require extracting parameters from each protocol's live contracts and computing profitability envelopes. --> Relevant Notes: - [[self-liquidation via flash loan is profitable when borrowed funds trigger and profit from the borrowers own liquidation before repaying the flash loan]]: the concrete pattern this question analyzes at the cross-protocol level Topics: - [[exploit-analyses]] - [[protocol-mechanics]]