The lending market does not require a principal-agent model

Core Viewpoint

BlockBeats

2025-12-10 22:24:11

Collection

When matchmaking can be achieved through code, human intervention is no longer necessary.

Original Title: CLOB Lending: Markets Don't Need Curators

Original Author: @0xJaehaerys, Gelora Research

Original Compilation: EeeVee, SpecialistXBT, BlockBeats

Editor's Note: After the collapse of Stream Finance and USDX, the DeFi community is undergoing a painful disillusionment. The "Curator" model introduced by protocols like Morpho and Euler was intended to address liquidity fragmentation, but inadvertently reintroduced the moral hazard of "humans" back onto the chain. The author points out that current lending protocols mistakenly bundle "risk definition" with "order matching." By drawing on the order book model from traditional finance, this article constructs a new paradigm that operates without curators and is automatically routed by algorithms.

The Evolutionary Logic of the Lending Market

Reviewing the evolutionary history of on-chain trading can provide insights into understanding the lending market.

Constant function AMMs (like Uniswap) solved a fundamental problem: how to create a market without active market makers? The answer is to use invariant functions to preset the "shape" of liquidity. Liquidity providers agree in advance on a set of strategies, which are automatically executed by the protocol.

This works well in trading because the transactions are relatively simple: buyers and sellers meet at a certain price. But lending is much more complex. A loan involves multiple dimensions:

Interest Rate
Collateral Type
Loan-to-Value Ratio (LTV)
Term (Fixed vs. Variable)
Liquidation Mechanism

Matching loans requires satisfying constraints across all of the above dimensions simultaneously.

Early DeFi lending directly adopted solutions similar to AMMs. Protocols like Compound and Aave preset interest rate curves, allowing lenders to join a shared liquidity pool. This enables the lending market to operate even without active lenders.

However, this analogy has a fatal flaw. In DEX trading, the shape of the constant function curve affects execution quality (slippage, depth); in lending, the shape of the interest rate curve directly determines risk. When all lenders share a pool, they also share the risk of all collateral accepted by that pool. Lenders cannot express their desire to only take on specific risks.

In trading, the order book solves this problem: it allows market makers to define their own "curve shapes." Each market maker quotes prices at levels they are comfortable with, and the order book aggregates these quotes into a unified market, but each market maker still controls their own risk exposure.

Can lending adopt the same approach? A project called Avon attempts to answer this question.

The Liquidity Fragmentation Dilemma

To give lenders control, the first attempt in DeFi was market isolation.

Protocols like Morpho Blue and Euler allow anyone to create lending markets with specific parameters: designated collateral, borrowed assets, fixed liquidation LTV, and interest rate curves. Lenders deposit into markets that align with their risk preferences. A bad debt in one market will never affect another market.

This is perfect for lenders, as they achieve the desired risk isolation.

But for borrowers, this creates fragmentation.

Taking ETH-USDC lending as an example, there may be a dozen different markets:

Market B: $3 million liquidity, 86% LTV, 5.1% interest rate
Market C: $2 million liquidity, 91% LTV, 6.8% interest rate
…and 9 other markets with lower liquidity

A user wanting to borrow $8 million cannot find satisfaction from a single market. They must manually compare prices, execute multiple transactions, manage fragmented positions, and track different liquidation thresholds. The theoretical optimal solution requires splitting the loan across more than four markets.

In reality, no one does this. Borrowers typically choose only one market. Funds are underutilized in fragmented pools.

Market risk isolation solves the lender's problem but creates issues for borrowers.

Limitations of the Curated Vault

The curated vault model attempts to bridge this gap.

The idea is that funds are managed by professional curators. Lenders deposit into the vault, and curators allocate funds to the underlying markets, optimizing returns and managing risks. Borrowers still face fragmented markets, but at least lenders do not have to manually adjust their positions.

This helps lenders who want to "take it easy," but it reintroduces something that DeFi originally aimed to eliminate: discretion.

Curators decide which markets receive funding and can reallocate at any time. Lenders' risk exposure changes with the curator's decisions, which are unpredictable and uncontrollable. As one Twitter user put it: "Curators are PvPing with borrowers, but borrowers don't even know they're being harvested."

This asymmetry is evident not only in strategy but also in the accuracy of the underlying interface. Morpho's UI sometimes shows "there is $3 million available liquidity," but in practice, low-interest funds are scarce, with most funds located in high-interest ranges.

When liquidity coordination relies on human decisions, transparency suffers.

Fund allocators adjust market liquidity according to their schedules rather than the immediate needs of the market. The vault attempts to solve the borrower's fragmentation issue through "rebalancing," but rebalancing incurs gas fees, relies on the curator's willingness, and is often delayed. Borrowers still face suboptimal rates.

Separating Risk from Matching

Lending protocols confuse two distinct modules.

User-defined risk: Different lenders have varying views on collateral quality and leverage.

The protocol's method of matching loans: This is mechanical. It does not require the user's subjective judgment, only efficient routing.

The fund pool model bundles both, causing lenders to lose control.

The isolated pool model separates risk definition but sacrifices matching, forcing borrowers to manually seek the optimal path.

The curated vault model reintroduces matching through the curator's role but introduces a trust assumption for the curator.

Can matching be automated without introducing discretion (human intervention)?

The order book in trading achieves this. Market makers define quotes, the order book aggregates depth, and matching is deterministic (best price first). No one decides where orders go; the mechanism determines everything.

CLOB lending applies the same principles to the credit market:

Lenders define risk through isolated strategies.
Strategies publish quotes to a shared order book.
Borrowers interact with unified liquidity.
Matching occurs automatically without curator intervention.

Risk remains with the lenders, and coordination becomes mechanical. No trust in third parties is required at any stage.

Two-Tier Architecture

Avon achieves order book lending through two unique tiers.

Strategy Layer

A "strategy" is an isolated lending market with fixed parameters.

Strategy creators define the following parameters: collateral/borrowed assets, liquidation LTV, interest rate curve, oracle, liquidation mechanism.

Once deployed, the shape of the interest rate curve cannot be changed. Lenders know the rules exactly before depositing.

Funds never move between strategies.

If you deposit into Strategy A, your money stays in Strategy A until you withdraw. No curators, no rebalancing, no sudden changes in risk exposure.

While there are still people (strategy managers) setting parameters, they are fundamentally different from curators: curators are fund allocators (deciding where the money goes), while strategy managers are true risk managers (defining rules but not moving money), similar to Aave DAO. The decision-making power of fund allocation always remains with the lenders.

How does the system adapt to market changes? Through competition, not parameter modification. If the risk-free rate spikes, it forces old strategies to be phased out (funds flowing out) and new strategies to be created (funds flowing in). "Discretion" shifts from "where should the funds go" (curator decision) to "which strategy should I choose" (lender decision).

Matching Layer

Strategies do not directly serve borrowers but publish quotes to the shared order book.

The order book aggregates quotes from all strategies into a unified view. Borrowers see the combined depth of all strategies that accept their collateral.

When a borrower places an order, the matching engine will:

Filter quotes by compatibility (collateral type, LTV requirements).
Sort by interest rate.
Execute starting from the cheapest.

It settles in an atomic transaction.

If a strategy can fulfill the entire order, it does so; if not, the order is automatically split across multiple strategies. Borrowers only perceive one transaction.

Important Note: The order book only reads the strategy state and cannot modify it. It is solely responsible for coordinating access and has no authority to allocate capital.

The Gospel of RWA

DeFi has faced structural contradictions in institutional adoption: compliance requirements isolate, but isolation stifles liquidity.

Aave Arc attempted a "walled garden" model, where compliant participants have their own pools. The result is shallow liquidity and poor rates. Aave Horizon tried a "semi-open" model (RWA issuers require KYC, but lending is permissionless), which is progress, but institutional borrowers still cannot access Aave's main pool of $32 billion in liquidity. Some projects explored permissioned rollups. KYC processes are completed at the infrastructure level. This approach suits certain use cases but leads to fragmented liquidity at the network layer. Compliant users on Chain A cannot access liquidity on Chain B.

The order book model offers a third way.

The strategy layer can implement any access control (KYC, geographic restrictions, accredited investor checks). The matching engine is only responsible for matching.

If a compliant strategy and a permissionless strategy both offer compatible terms, they can simultaneously fill the same loan.

Imagine a corporate treasury collateralizing tokenized government bonds to borrow $100 million:

$30 million from a strategy requiring institutional KYC (pension fund LP)
$20 million from a strategy requiring accredited investor certification (family office LP)
$50 million from a fully permissionless strategy (retail LP)

Funds are never mixed at the source, institutions remain compliant, but liquidity is unified globally. This breaks the deadlock of "compliance equals isolation."

The Mechanism of Multi-Dimensional Matching

The order book matches on one dimension: price. The highest bid matches the lowest ask.

The lending order book must match across multiple dimensions simultaneously:

Interest Rate: Must be below the borrower's acceptable maximum.
LTV: The borrower's collateral ratio must meet strategy requirements.
Asset Compatibility: Currency matching.
Liquidity: The market must have sufficient liquidity.

Borrowers can provide more collateral (lower LTV) or accept higher interest rates to match more strategies. The engine will find the cheapest path within this constraint space.

For large borrowers, one point to note is that in Aave, $1 billion in liquidity is a single fund pool. In order book lending, $1 billion may be spread across hundreds of strategies. A $100 million loan will quickly consume the entire order book, starting from the cheapest strategies and gradually filling up to the most expensive. Slippage is apparent.

Pool-based systems also experience slippage, just in a different form: a surge in utilization will push up interest rates. The difference lies in transparency. In the order book, slippage is visible in advance. In pool-based systems, slippage only becomes apparent after the transaction is executed.

Floating Rates and Re-quoting

DeFi lending uses floating rates. As utilization changes, interest rates also change.

This brings a synchronization challenge: if the strategy's utilization changes but the quotes on the order book are not updated, borrowers may transact at incorrect prices.

Solution: Continuous re-quoting.

Once a strategy's state changes, new quotes are immediately published to the order book. This requires extremely high infrastructure performance:

Very fast block times.
Very low transaction costs.
Atomic state reading.

This is also why Avon chose to build on MegaETH. On the Ethereum mainnet, this architecture is unfeasible due to high gas fees.

Existing frictions:

If market interest rates change but the strategy's fixed curve does not adapt, a "dead zone" may occur—borrowers find the rates too high to borrow, and lenders earn nothing. In Aave, the curve adjusts automatically, while in CLOB mode, this requires lenders to manually withdraw and migrate to new strategies. This is the price paid for gaining control.

Multi-Strategy Position Management

When a loan is filled by multiple strategies, the borrower effectively holds a multi-strategy position.

Although it appears as a single loan on the interface, the underlying components are independent:

Independent Interest Rates: The interest rate of component A may rise due to increased utilization of strategy A, while component B remains unchanged.
Independent Health Ratios: When the price of the asset drops, components with stricter LTV limits will be liquidated first. You won’t face a total liquidation at once but will experience a series of partial liquidations like being "eroded."

To simplify the experience, Avon offers unified position management (one-click collateral addition, automatic weighted allocation) and one-click refinancing (automatically borrowing new to pay old through flash loans, always locking in the market's best rates).

Conclusion

DeFi lending has gone through several stages:

Pool Protocols (Pooled): Provided depth for borrowers but stripped control from lenders.
Isolated Markets (Isolated): Gave control to lenders but fragmented the borrowing experience.
Curated Vaults (Vaults): Attempted to bridge the two but introduced the risk of human decision-making.

Order Book Lending (CLOB): Decouples the above models. The power to define risk returns to lenders, and matching is achieved through the order book engine.

This design principle is clear: when matching can be implemented through code, human intervention is no longer needed. Markets can self-regulate.