Liquidation Parameters and Risk Tuning

Synthetix V3's liquidation engine is a critical solvency backstop, designed to close under-collateralized positions before they impose losses on the shared debt pool. The engine's behavior is not static; it is governed by a set of on-chain parameters that can be adjusted via Synthetix Configuration Change Proposals (SCCPs). These parameters directly control the system's risk tolerance, the user experience during volatility, and the economic incentives for keepers to perform timely liquidations.

The core tunable parameters include the minimum collateralization ratio (c-ratio), the liquidation reward paid to keepers, and the self-liquidation window. The c-ratio defines the threshold at which a position becomes eligible for liquidation, balancing capital efficiency against systemic risk. The liquidation reward, often a percentage of the seized collateral, must be calibrated to ensure keeper participation is profitable enough to guarantee a robust network of liquidators, especially during gas spikes or network congestion. The self-liquidation window allows a position owner to close their own position after being flagged, avoiding the keeper penalty, which adds a user-protection layer to the process.

Tuning these parameters is a high-stakes governance action. Setting a c-ratio too low increases the risk of bad debt accrual during a sharp market downturn, while setting it too high makes capital usage inefficient for stakers. Similarly, an insufficient liquidation reward can lead to a keeper network failure, leaving the system without a liquidator precisely when one is needed most. Protocol operators, risk teams, and governance participants must model the downstream effects of any proposed change on market solvency, keeper economics, and user protection before an SCCP reaches an on-chain vote. Chainscore Labs provides independent impact assessments and parameter-change reviews to help teams validate these critical decisions.

Synthetix's liquidation engine relies on a precise set of on-chain parameters that define when a position becomes under-collateralized and how it is resolved. The core parameters include the liquidation ratio (the minimum collateral-to-debt threshold before a position is flagged), the liquidation penalty (a fee levied on the liquidated collateral to incentivize keepers), and the self-liquidation window (a grace period allowing users to restore their position's health before a third-party keeper can act). These values are not static; they are tuned per collateral type to reflect asset-specific volatility and liquidity profiles, directly linking governance decisions to the protocol's real-time solvency.

The interaction between these parameters creates a dynamic risk surface. A low liquidation ratio increases capital efficiency for users but shrinks the safety buffer for the system, raising the risk of bad debt during a flash crash. A high liquidation penalty can attract a robust keeper network but may disincentivize self-liquidation, leading to worse user outcomes. The self-liquidation window must be long enough for users to react to market moves but short enough to prevent a position from accruing unrecoverable losses. Governance participants must model these trade-offs, understanding that a parameter change on one collateral type can shift keeper economics and risk exposure across the entire debt pool.

For operators and integrators, these parameters are the critical signals for monitoring system health. A sudden governance proposal to tighten liquidation ratios across multiple collateral types may indicate an anticipated rise in market volatility. A failure of keepers to liquidate positions promptly, even when parameters are met, signals a breakdown in keeper network economics that requires immediate investigation. Chainscore Labs helps risk teams and protocol operators model the impact of parameter changes, stress-test liquidation cascades, and build the monitoring infrastructure needed to verify that the liquidation engine behaves as designed under all market conditions.