The Volatility Harvest: How DeFi Protocols Transformed Liquidation into a Revenue Engine

Decentralized finance dispenses with this traditional architecture. In its place, DeFi utilizes over-collateralized borrowing, a mechanism requiring users to deposit assets exceeding the value of their loan. These assets are held in non-custodial smart contracts, ensuring the lender is protected without the need for personal identification or credit scores.

Lending protocols manage this risk by assigning a Loan-to-Value (LTV) ratio to each deposit. This ratio defines the maximum borrowing capacity, with the remaining gap acting as a vital buffer against market volatility. While this design enables permissionless lending to function at scale, it also gives rise to a 'secondary economy' of liquidators and arbitrageurs. This ecosystem becomes most visible, and most critical, precisely when market prices turn against borrowers, triggering automated safeguards to preserve the protocol's solvency.

Sometimes, sharp market moves do not unfold gradually enough for borrowers to react in an orderly fashion. The buffer that appeared comfortable at origination can evaporate within a single candle.

When this occurs, the protocol faces a binary choice. It can either absorb the deteriorating collateral ratio and hope prices recover, or it can act immediately by liquidating the collateral to protect the system. Every major lending protocol has chosen the latter. These platforms utilize automated liquidation engines that trigger the moment a collateral value breaches a liquidation threshold or maintenance margin.

Lending protocols’ liquidation engine 

When a position becomes liquidatable, the protocol does not dispose of the collateral itself. Instead, it opens the position to external actors known as liquidators. These participants are invited to repay some or all of the borrower's outstanding debt. In exchange, they receive a portion of the collateral, which is delivered at a discount to the current market price.

In practice, the vast majority of liquidation activity is carried out by MEV (Maximal Extractable Value) bots. These are automated systems operated by specialized teams that monitor the blockchain state in real time. They detect liquidatable positions the moment an oracle updates a price feed, and submit transactions optimized for rapid execution.

These bots typically use flash loans to avoid tying up capital. They borrow the repayment amount within the same transaction block, execute the liquidation, and collect the discounted collateral. They then sell the assets, repay the flash loan and pocket the spread. This entire process occurs atomically, meaning the liquidator carries no net capital at risk.

DeFi lending protocols have built business models in which market volatility is not a risk to be hedged but a direct revenue driver. This dynamic reflects a broader rule of thumb across the ecosystem. Protocol revenues are fundamentally a derivative of market activity. Fees on trading, borrowing, liquidation penalties and interest spreads all expand when volatility spikes. For instance, the protocols covered in this analysis recorded peak weekly revenues around the sharp market dislocations of 10 Oct 2025 and the first week of February 2026.

(Source: Token Terminal)

The system remains only as robust as the incentive calibration that underpins it. Protocols define the specific parameters, including the threshold, the discount, and eligible collateral, and allow the market to clear. If these incentives are sufficient, liquidators compete to close underwater positions rapidly to maintain protocol solvency. If the incentives are too thin, liquidations stall, bad debt accumulates, and the system becomes exposed.

Actual capital risk only materializes under a specific and simultaneous convergence of conditions. This occurs when collateral prices move faster and further than any liquidator can act on, within a market too illiquid to absorb the resulting sell pressure. Protocol buffers are engineered to prevent this. The gap between the LTV ceiling and the liquidation threshold, combined with the liquidation bonus itself, is theoretically designed to absorb most mark-to-market losses well before that point.

Still, the mechanics of liquidation are not uniform across the ecosystem. Each protocol makes distinct architectural choices that reflect different priorities. These include revenue capture, borrower protection, capital efficiency and systemic solvency.

AAVE and Chainlink SVR: from revenue leakage to revenue capture 

Aave v3 operates a liquidation system defined by two primary parameters. The first is the liquidation threshold, which represents the collateralization ratio below which a position becomes eligible for liquidation. The second is the liquidation bonus, which is the specific discount offered to liquidators on seized collateral as compensation for closing the position.

The classic model fails to address a structural value leak that went entirely unrecaptured until last year. This is known as Oracle Extractable Value (OEV). Every time a Chainlink price feed updates and renders a position liquidatable, the price delta between the old oracle price and the new one creates a specific window of extractable value.

In March 2025, Aave integrated Chainlink Smart Value Recapture (SVR) on the Ethereum mainnet. This system ensures that updated prices are transmitted simultaneously through the public mempool and a private channel. Searchers then bid for the right to backrun the oracle update with a liquidation transaction in the same block.

(Source: Dune Analytics)

Since the Chainlink SVR launch, the system has processed $683mn in total liquidation volume. Across these transactions, $32.2mn in liquidation bonuses were paid out to liquidators.

The more consequential figure is the $18.2mn in SVR revenue recaptured from non-toxic information asymmetry at the oracle level. All of this revenue was generated during windows of market stress. Of that total, $11.83mn accrued to Aave and $6.4mn to Chainlink. This represents value that previously leaked entirely to external MEV actors and returned nothing to the protocols generating it.

To put those figures in context, SVR revenue represents approximately 19.3% of Aave's total revenue and 8.7% of Chainlink's total revenue over the same period. This marks a material contribution from a mechanism that did not exist prior to the March 2025 integration.

According to TokenLogic, the system currently operates at a ~73% average recapture rate. This means nearly three-quarters of the extractable value generated at each oracle update is successfully redirected to the protocol rather than absorbed by the open MEV market. The remaining 27% continues to leak, which sets a clear ceiling for what further optimization of the SVR infrastructure could still recover.

Sky, the Auction-based liquidation and the Surplus Buffer 

Sky takes a fundamentally different approach. Rather than incentivizing external liquidators through a fixed discount on seized collateral, the protocol auctions the collateral directly. This is handled through an internal Dutch auction mechanism. In this model, the price starts high and decreases over time until a buyer, referred to as a Keeper, submits a bid sufficient to cover the vault's outstanding debt plus the liquidation penalty.

The liquidation penalty is set per collateral type by governance. For standard ETH vaults, this has historically been 13% above the outstanding debt. This structure ensures that the system is repaid in full while any remaining collateral value is returned to the original borrower, theoretically minimizing the loss for the user compared to fixed-discount models.

(Source: Dune Analytics)

The proceeds of a successful collateral auction flow directly into the Sky Protocol Buffer. This reserve is utilized for SKY token buybacks. Across the full observation period, Sky generated $20.9mn in liquidation revenue. Of this total, $4.1mn accrued in 2024 and $16.5mn in 2025.

April 2025 alone produced $11.5mn. This represents half of the total revenue over the past two years and nearly 70% of all 2025 liquidation income. This surge was driven by the sharp market correction on Liberation Day. It serves as a precise illustration of the structural lumpiness of liquidation income. Revenue remains negligible across extended periods of low volatility but becomes overwhelmingly concentrated into a handful of stress episodes.

Hyperliquid’s liquidation waterfall

Lending protocols operate on over-collateralized positions where liquidation is a scheduled response to a deteriorating collateral ratio. Perpetual futures introduce a structurally different risk profile. Leverage means positions can move from healthy to insolvent faster than any buffer can absorb. This requires a more aggressive liquidation architecture.

Hyperliquid handles position risk through a three-tier waterfall. Each layer is activated only when the one before it fails. This sequential approach is designed to contain losses within the specific account before they can impact the broader protocol's solvency.

The first tier consists of standard order book liquidation. This is triggered when a trader's account equity falls below the maintenance margin. Notably, Hyperliquid charges no clearance fee at this stage. Any residual margin after the position is closed is returned to the trader.

When order book liquidity is insufficient to fully close the position, the Hyperliquidity Provider (HLP) vault steps in. This is a community-funded backstop that absorbs the remaining position and unwinds it passively through the book over time. At this point, the trader's maintenance margin is not returned. It is instead retained by the vault as a buffer to ensure the takeover remains economically viable. In normal conditions, backstop liquidations are profitable for HLP. The vault acquires distressed exposure below market value and recovers it as prices stabilize.

The third and final tier is Auto-Deleveraging (ADL). This mechanism is summoned only when both the order book and the HLP vault have failed to absorb the mark-to-market selling pressure. At this stage, the protocol force-closes positions against the most profitable and most leveraged counterparties on the opposite side of the trade.

(Source: Hyperliquid HLP)

Since its launch in May 2023, HLP has generated approximately $138.3mn in cumulative profit for liquidity providers through market-making and liquidation activity. The 10 Oct bloodbath yielded $41.4mn to the protocol in a single week, marking the highest weekly print since inception. The recent drawdown added a further $18.8mn, once again capitalizing on volatility spikes.

The model is not without vulnerabilities. In March 2025, a single wallet opened a $200mn long ETH position at near-maximum leverage. The trader then deliberately withdrew a large portion of the margin, forcing the position into backstop liquidation. HLP was obliged to absorb the position at an unfavorable entry, which resulted in a $4mn loss to the vault. Meanwhile, the trader exited with an estimated $1.8mn in profit.

The episode exposed a structural weakness in how the protocol handled large leveraged positions and deliberate margin withdrawal as a liquidation trigger. Hyperliquid responded by cutting maximum leverage and raising effective maintenance margin requirements for large orders to reduce the attack surface.