Traditional lending systems are built around a simple assumption. At some point, a position reaches a level where action has to be taken. A user deposits collateral and borrows against it. The position remains healthy as long as the collateral value stays above a certain threshold. Once that threshold is crossed, liquidation begins. This model has powered a large part of DeFi lending. Money markets like @aave work by connecting borrowers with liquidity supplied by other users. Users deposit assets into a pool, and borrowers take liquidity from that pool. CDP systems like $crvUSD follow a different structure. Users provide collateral and mint a new stablecoin against their position, creating additional stablecoin supply that can be used throughout the ecosystem. The structures are different, but both systems face the same fundamental question: What happens when markets move quickly? A sharp price drop can push many positions toward liquidation at the same time. The protocol needs to protect itself, liquidators step in, collateral is sold, and the borrower's position changes immediately. The mechanism works exactly as designed. The challenge is the transition itself. A position can move from healthy to liquidated because of a short period of volatility. The market moves continuously, but liquidation usually happens at a specific moment. That gap is where Curve focused its research. Instead of only optimizing the existing liquidation process, Curve redesigned how collateral reacts to changing market conditions. This idea became LLAMMA. Lending-Liquidating AMM Algorithm. The core concept is simple. Collateral does not move directly from "safe" to "liquidated". LLAMMA creates a gradual process where collateral can shift between assets as the market moves. This is known as a soft liquidation. When collateral prices move against the borrower, LLAMMA gradually converts parts of the collateral into the borrowed stablecoin. If the market recovers, the process can reverse and the position can move back toward its original state. The position continues to exist. The system adapts. The key innovation behind this design is the use of price bands. Instead of one liquidation point, collateral is distributed across different price ranges. As the market moves through these ranges, the AMM continuously adjusts the position. Risk becomes something that changes over time instead of something that suddenly switches on. This changes the relationship between borrowers, protocols, and liquidators. The goal is to manage collateral throughout the entire movement of the market. This follows the same design philosophy that has shaped @CurveFinance from the beginning. StableSwap redesigned how stable assets trade. veCRV redesigned how governance power is created. Gauges redesigned how liquidity incentives are allocated. LLAMMA redesigned how collateral risk can be managed. The mechanism behind LLAMMA is what gives crvUSD a fundamentally different approach to collateral management. The next step is looking at a real example. What actually happens when someone opens a crvUSD loan, the collateral price falls, and the position moves through different price bands?