How DeFi Is Trying to Solve Undercollateralized Lending

Undercollateralized lending is one of DeFi's most persistent unsolved problems. The potential is enormous—it's how traditional finance actually works—but every approach comes with tradeoffs. Some rely on trust. Some constrain what borrowers can do. Some have already failed.

Here's how the landscape breaks down.

Reputation and Credit Scores

The most intuitive approach: build a credit score for DeFi. If traditional finance can assess creditworthiness based on payment history, why can't DeFi do the same with on-chain data?

Several protocols have tried. Spectral Finance calculates a "MACRO score" using on-chain transaction history—payment patterns, liquidation history, amounts borrowed and repaid. ARCx built a DeFi Credit Score that lets borrowers with strong track records access higher loan-to-value ratios. RociFi combines on-chain data with decentralized identity signals (Twitter, GitHub, DAO participation) to generate a non-fungible credit score.

The appeal is obvious: reward good behavior, penalize bad actors, and let capital efficiency improve for borrowers who've earned it.

The problem: reputation is thin in crypto. A wallet with a perfect repayment history can be abandoned the moment it's profitable to default. Unlike traditional credit scores tied to legal identity, on-chain reputation doesn't follow bad actors—they just create a new wallet. And borrowers with long histories may simply be farming a score before executing a larger default.

Some protocols address this by linking scores to identity verification, but that reintroduces the centralization and KYC requirements that DeFi was designed to avoid.

Verdict: Promising in theory, but pseudonymity makes reputation fragile. Works best when combined with other mechanisms.

Delegate-Based Lending

This model dominated institutional DeFi lending until 2022. Protocols like Maple Finance, TrueFi, Clearpool, and Goldfinch use a similar structure: pool delegates or backers assess borrowers, approve loans, and take responsibility for credit decisions.

Lenders deposit funds into pools. Delegates evaluate institutional borrowers—trading firms, market makers, real-world businesses—and approve loans at negotiated terms. The delegate often posts first-loss capital, creating some skin in the game.

Maple focused on crypto-native institutions like trading firms. TrueFi required token-holder approval for loans. Clearpool differentiated with variable interest rates. Goldfinch targeted emerging market businesses, requiring off-chain collateral and legal agreements.

Pre-2022, default rates were low: TrueFi at 0.2%, Clearpool at 0%, Goldfinch at 0%. Then FTX collapsed.

Orthogonal Trading, a borrower on Maple, had claimed minimal FTX exposure. When they defaulted on $36 million in loans, depositors faced up to 80% losses. The delegate system failed because delegates couldn't verify borrower claims—they were trusting representations that turned out to be false.

Maple's response was telling: every pool launched after 2022 required overcollateralization. The leading undercollateralized lending protocol essentially abandoned the model.

Verdict: Works in calm markets, fails under stress. Delegates become single points of failure, and their due diligence can't catch fraud or fast-moving contagion.

Credit Delegation

Aave offers a different approach: let depositors delegate their borrowing power to others. If you've deposited collateral but don't need to borrow against it, you can authorize someone else to borrow using your credit line.

From the borrower's perspective, this is undercollateralized—they're borrowing without posting their own collateral. From the protocol's perspective, it's still fully collateralized—the delegator's deposits back the loan.

The problem: this just relocates trust. The delegator must trust the borrower to repay. If the borrower defaults, the delegator's collateral gets liquidated. The protocol is protected, but the delegator bears all the risk.

Credit delegation works for trusted relationships—a fund lending to its own trading desk, or institutions with legal agreements in place. It doesn't solve the general problem of enabling undercollateralized borrowing at scale.

Verdict: Useful for specific use cases, but it's trust-based lending with extra steps. Doesn't scale beyond existing relationships.

Structural Constraints with First-Loss Capital

Flux takes a fundamentally different approach: instead of trusting people, constrain what they can do and make them post capital at risk.

Managers on Flux can only interact with whitelisted assets and protocols. They can't withdraw borrowed funds to arbitrary wallets or deploy them into unvetted strategies. But within those rails, they have real flexibility—trading perps, providing liquidity, running arbitrage across venues.

The critical addition is the manager's bond. Every manager must post their own capital as first-loss protection—typically 20% of what they borrow. This bond sits in the vault, absorbing losses before LP funds are touched.

If a manager's position becomes unhealthy, anyone can liquidate it. The liquidator unwinds the positions, repays the debt, and the manager's bond covers any shortfall. Only in extreme cases—where losses exceed the bond—do LPs face bad debt.

This creates different incentives than delegate-based systems. Delegates on Maple had some first-loss exposure, but they were also earning fees for approving loans. A Flux manager's bond isn't earning yield while sitting idle—it's pure downside protection. The manager is incentivized to run profitable strategies, not just to approve borrowers.

Solving liquidity fragmentation

Flux's omni-chain architecture addresses another problem that has limited DeFi lending: fragmented liquidity across chains. LPs can deposit on, e.g., Ethereum L1, and managers can borrow that capital to execute strategies on Solana, Hyperliquid, or other chains. The vault doesn't need separate pools on every chain—capital flows where opportunities are, while LPs maintain a single deposit position.

This matters because opportunities aren't confined to one chain. A manager might spot yield on Solana, an arbitrage on Hyperliquid, and a liquidity provision opportunity on Arbitrum—all in the same week. Without omni-chain infrastructure, they'd need to source capital separately on each chain, or LPs would need to fragment their deposits across multiple venues. Flux unifies the capital layer while letting execution happen wherever the returns are.

Verdict: Combines structural safety with meaningful first-loss capital and cross-chain flexibility. Doesn't require trusting delegates or relying on fragile reputation scores. The bond structure aligns incentives: managers only profit if their strategies work, and they lose real capital if they don't.

Where This Leaves Us

No solution is perfect. Reputation scores struggle with pseudonymity. Delegate models fail when trust is misplaced. Credit delegation just moves risk around.

Flux's approach—structural constraints, first-loss capital, and omni-chain deployment—doesn't eliminate risk, but it changes who bears it and how it's managed. The manager has skin in the game. The protocol has rails that prevent the worst outcomes. LPs have protection that doesn't depend on trusting the right people, and their capital can access opportunities across the entire crypto ecosystem from a single deposit.

Undercollateralized lending in DeFi isn't a solved problem. But the solution probably looks less like traditional credit assessment and more like structural incentive design. Trust humans less. Trust mechanisms more.