Most loans follow you. If you stop paying a bank, it can chase you for the rest. A crypto loan usually does not work that way. You can stop paying, give up the collateral, and owe nothing more.

That feature has a name: non-recourse. It is easy to read past it as fine print. It is actually the key to the whole design. Once you see what a non-recourse loan is made of, the odd numbers in crypto lending make sense. That includes the 50% starting LTVs, the liquidation thresholds, and the lenders who want far more collateral than the loan is worth.

What non-recourse means in a crypto loan

In a non-recourse loan, the lender’s claim stops at the collateral. Nothing else you own is at risk.

Researchers at the Bank of Canada and several universities describe DeFi lending this way. They say it typically involves linear, non-recourse debt contracts that feature overcollateralization as the only risk control. The reason is practical. Contract terms are pre-programmed and cannot depend on soft information such as news or sentiment. A smart contract cannot sue you. It does not know your name. It can only hold the collateral and sell it.

Centralized lenders look different on the surface. Coinbase’s loans, for example, need no credit check because the collateral secures the debt. The logic is the same. The asset is the loan’s security, and the borrower’s identity barely matters.

The connection most explainers skip

Here is the non-obvious part. A non-recourse loan is really two things bundled together. One is a plain loan. The other is a put option on your collateral, and you own it.

Think about your choices at repayment. If the collateral is worth more than the debt, you repay and take it back. If it is worth less, you hand it over and keep the cash you borrowed. You get to pick whichever is better for you. That is what a put option is: the right to sell something at a fixed price, which here is the amount you owe.

This is not a new idea. Finance has used it for corporate debt since 1974. Robert Merton’s insight was that a firm’s equity is a call option on its assets. The reason is limited liability. Shareholders get unlimited upside if the firm thrives and a hard floor if it fails: they lose their stake and nothing more. The lender’s side follows. Risky debt equals risk-free debt minus the value of the implicit put option that limited liability gives shareholders.

A crypto loan is the same structure in a purer form. Limited liability applies to the whole loan, and the “firm” is a single pile of collateral. In Merton’s model, shareholders hand the keys to the bondholders instead of topping up the shortfall. A crypto borrower hands over the coins.

Who wrote the put? The lender did

Someone has to be on the other side of that option. In a crypto loan, it is the lender. The lender has sold you a put, and you did not pay a separate premium for it.

Corporate lenders can price this risk. They read financial statements, set covenants and charge higher rates to riskier firms. A pool of anonymous lenders on a blockchain cannot do most of that. The Bank of Canada paper notes that lenders cannot control or easily monitor the composition of the underlying collateral pool. So the main lever left is the strike of the put. The lender sets it so far from the current price that it is unlikely to pay out.

Some designs make the premium explicit. One academic proposal for “zero-liquidation loans” splits the borrower’s position into a loan and an option. In that structure, the interest is essentially the put option premium. Ordinary crypto loans bury the same price inside their terms.

Reading LTV and liquidation thresholds as option terms

With that frame, the usual jargon becomes plain.

Loan-to-value is how far out of the money the put starts

If you borrow against 75% of your collateral’s value, the put is struck at 75% of today’s price. The collateral has to fall by a quarter before walking away would pay. Aave’s documentation gives this very example: a 75% LTV allows borrowing 75% of the collateral’s value. Lenders also adjust this by asset. In one Aave risk framework, the most volatile assets were set at an LTV of 35% to 40% and a 65% liquidation threshold.

That fits option logic. A put is worth more when the underlying asset is more volatile. Merton’s framework says the same thing from the other side: higher asset volatility raises equity value and default probability. A lender writing puts on a jumpy asset wants the strike much further away.

The liquidation threshold is the lender’s early-exit trigger

A bank’s put only matters at maturity. A crypto lender cannot wait that long, because prices move every second and no one can pursue you later. So the contract adds a trigger. Liquidation happens when a borrower’s health factor falls below 1, meaning the collateral no longer sufficiently covers the debt. The formula is simple. Health factor equals total collateral value times the weighted liquidation threshold, divided by total borrow value.

Take Aave’s own example. A user supplies $10,000 in ETH with an 80% liquidation threshold and borrows $6,000, giving a health factor of 1.333. By my arithmetic, the collateral would have to fall 25% before that position can be liquidated.

In option terms, the trigger forces the exercise early. The protocol sells the collateral while it is still worth more than the debt. The liquidator repays the debt and receives equivalent value plus a liquidation bonus from the borrower’s collateral. The gap between the threshold and 100% pays for that bonus and for slippage. It is the lender’s margin for error.

Real products show the same pattern. Coinbase opens loans at a collateral ratio of at least 133%, a maximum LTV of about 75%, and liquidates if LTV reaches 86%. That leaves roughly 13% of price movement between the maximum borrow and the trigger, by my calculation. Ledn requires an initial LTV of 50% and starts warning borrowers at 70%. The two lenders chose different cushions. Both are trying to keep the put out of the money.

What happens when the put pays out

Liquidation is meant to make sure the lender is repaid before the put goes deep into the money. Sometimes it fails, and November 2022 on Aave showed how.

One trader built a huge position. Over ten days from November 14, he amassed a loan of 92 million CRV, about $39 million at the time, using $50 million in USDC as collateral. The collateral was a stablecoin and the debt was a volatile token. So the price risk ran the other way: the put gained value when CRV rose. Once the borrowed CRV rose above 88% of the USDC collateral, liquidators stepped in to sell the USDC, buy back CRV and close the position. (The research paper cited below puts the threshold at 89%.)

CRV was thinly traded on-chain. The position was closed over 50 minutes in 385 small liquidations, leaving Aave with 2.64 million CRV, about $1.7 million, in bad debt. That is debt nobody will repay. The loss was small. Aave said it was less than 0.1% of borrows on the protocol. Reports put the figure between $1.6 million and $1.78 million.

The put was exercised, and the lender paid. Collateral did not cover the debt, and the borrower owed nothing further. In a recourse loan, the lender would have gone after the borrower. Here the contract gave no such claim.

Where the analogy gets debated

Treating a crypto loan as a loan plus a put is a model, and models leave things out. Three points are worth knowing.

Was that failure about volatility or design? Aave said its liquidations worked as designed but the position size left excess debt. It said the liquidations worked as designed but left “some excess debt.” A research paper argued differently. Its authors said the bad debt came from a flaw in the liquidation logic that triggered a toxic liquidation spiral, not from excess volatility. If they are right, the put’s price depends on how liquidation is built, not just on the asset’s volatility. That is a hard thing to price.

Borrowers are not pure option traders. A textbook option holder exercises whenever it pays. Most people borrowing against crypto want to keep the coins. Walking away costs them the asset they did not want to sell. For them, the put is a safety net more than a strategy. The Merton analogy also assumes the borrower is rational and the price moves smoothly. Neither holds in a crash.

Many loans have no end date, so the option never expires. Some loans have fixed terms. Others do not. Coinbase’s variable-rate loans have no monthly minimum payments or fixed repayment deadline. But that does not remove the risk. Falling collateral prices or accumulating interest can raise the LTV and expose a position to automatic liquidation. Interest quietly moves your strike closer to the current price over time.

What this means for you as a borrower

• Your downside is capped, but it is not small. You cannot owe more than you pledged. You can still lose all of it, and a forced sale usually happens at a bad price.
• You pay for the put through lost leverage. Borrowing only 50% of your collateral’s value is the lender’s price for the option you hold. It is not a rule to resent.
• Volatile collateral means a tighter box. The more an asset swings, the lower the LTV and threshold you should expect, and the less room you have.
• Watch the gap, not just the LTV. The distance between your current ratio and the liquidation trigger is the only buffer you control. Adding collateral or repaying part of the debt widens it.
• Non-recourse covers the loan, not the lender. Custody is a separate risk. Rehypothecation, where a lender re-lends your collateral, was a big reason customers of failed lenders lost their deposits in 2022. Check whether your collateral is held or re-lent.

Non-recourse is why crypto lenders ask for so much. They cannot chase you, so they hold the collateral instead. They also cannot judge your character, so they price in volatility. Once you see the put option inside the loan, those two rules look like one idea, and the rest of the terms follow from it.

Sources

• Health Factor & Liquidations (Aave Help)
• On the Fragility of DeFi Lending (Chiu, Ozdenoren, Yuan, Zhang; Bank of Canada working paper)
• The Merton Structural Model: Pricing Default as an Option on the Firm (Risk Hub)
• Merton Structural Model (MetricGate)
• Feature or Flaw? Aave Left With $1.7M in Bad Debt (Blockworks)
• Toxic Liquidation Spirals (Warmuz, Chaudhary, Pinna, arXiv)
• Mango Markets Exploiter Gets Liquidated But Leaves Aave With 'Excess Debt' (Unchained)
• Best Bitcoin Backed Loans (Bitcoin Magazine)
• 12 Best Crypto Loan Providers 2026 (CoinLedger)
• Aave: Health Factor and Collateral Risk (Blofin Academy)
• Zero-Liquidation Loans: A Structured Product Approach to DeFi Lending (ResearchGate)