Skip to content
Financial Context Wiki

DeFi lending loops

A lending loop repeatedly borrows against supplied collateral and resupplies the proceeds, creating leveraged exposure, recursive gross balances and liquidation risk without creating new net equity.

Updated

For educational purposes only; not investment advice. Lending loops use debt and can be liquidated. Rates, prices, liquidity, oracle data, protocol parameters and smart-contract behavior can change, and losses can exceed the apparent yield.

Direct answer

A DeFi lending loop is a sequence in which a user supplies collateral, borrows another asset against it, converts or wraps the borrowed asset into more eligible collateral, supplies that collateral, and repeats. The position has larger gross collateral and debt balances than the user’s initial capital. It creates leverage; it does not create new net equity.

The term describes a strategy, not one standardized protocol feature. Some interfaces or automation contracts bundle the steps, but every round still depends on the relevant market’s asset eligibility, loan-to-value limit, liquidation threshold, supply or borrow cap, oracle, interest model and available liquidity. Aave V3 allows supplied assets to remain interest-bearing while used as collateral; Compound III instead lends one base asset against supported collateral, and its supplied collateral assets do not earn interest.

Loops may target directional exposure, a supply-borrow rate spread, incentives, or repeated exposure to a staking or yield-bearing token. Borrowing and resupplying the same economic asset reduces price mismatch but does not remove interest, depeg, oracle, liquidity, contract or liquidation risk. Borrowing a stable asset to buy volatile collateral adds explicit leveraged price exposure.

Gross supply, protocol TVL and economic equity are different measures. Recursive balances can raise gross supplied value, and collateral chains can cause the same underlying capital to appear at several points across protocols. Whether an analytics product counts or nets those claims depends on its methodology; a headline TVL number alone does not measure unencumbered user wealth or loss capacity.

How it works

Let P be initial equity and r the fraction of each newly supplied amount that returns as new collateral after the borrowing limit, conversion slippage and fees, with 0 <= r < 1. After n borrow-and-resupply rounds, including the initial supply, the idealized balances are S_n = P * (1 - r^(n + 1)) / (1 - r) for gross supplied collateral and D_n = P * r * (1 - r^n) / (1 - r) for debt. Therefore S_n - D_n = P before price moves, interest and costs.

As n grows, S_n / P approaches 1 / (1 - r) and D_n / P approaches r / (1 - r). These are limits, not promised leverage. A wallet balance, cap, minimum borrow, rounding, fee, slippage or declining collateral factor can stop the sequence. Using the published maximum LTV as r also leaves little or no buffer for interest and price movement.

Collateral rules must be read from the exact deployment and block. On Aave, initial borrowing capacity is governed by LTV, while liquidation depends on the weighted liquidation threshold; its documented health factor is HF = collateral value * weighted liquidation threshold / borrow value, and liquidation eligibility begins below 1. E-mode supplies category-specific LTV, liquidation threshold, bonus and borrowing permissions. On Compound III, borrow collateral factors set initial capacity, separate higher liquidation collateral factors determine liquidatability, and collateral supplyCap values bound deposits.

Carry must be calculated on gross balances, not initial capital. A simplified annual estimate is net carry on equity = (S * supply rate - D * borrow rate + rewards - recurring costs) / P. This formula is only a ledger: rates vary with utilization, rewards can end or fall in price, and some collateral earns no protocol supply interest. One-time gas, swap spread, slippage, automation fees and exit costs must be amortized over the actual holding period.

Before entering, record the chain, market and contract addresses; collateral and debt assets; LTV or borrow factor; liquidation threshold or factor; caps; oracle and price units; current and stressed rates; swap route; approvals; and unwind transactions. Simulate both the entry and full exit. Monitor debt growth, health factor or liquidity, parameter changes, oracle status, available withdrawal liquidity and every dependency used by automation.

Worked example

Assume P = 10,000, r = 0.70, no costs and three borrow-and-resupply rounds. Gross collateral is S_3 = 10,000 + 7,000 + 4,900 + 3,430 = 25,330; debt is D_3 = 7,000 + 4,900 + 3,430 = 15,330; and idealized equity remains 25,330 - 15,330 = 10,000. Gross collateral exposure is 2.533x, below the infinite-round limit 1 / (1 - 0.70) = 3.333x.

Now assume the debt asset stays at its reference value, the collateral falls 10%, and the applicable liquidation threshold is 80%. Collateral becomes 22,797, equity becomes 22,797 - 15,330 = 7,467, a 25.33% equity loss before fees, and HF = 22,797 * 0.80 / 15,330 = 1.1897. The collateral price that makes this simplified position reach HF = 1 is 15,330 / (25,330 * 0.80) = 0.7564 of its entry price, or about a 24.36% decline. Accrued debt, parameter changes, oracle behavior and liquidation costs can move that boundary.

If the gross supply rate is 4% and the borrow rate is 6%, annual interest before rewards and costs is 25,330 * 0.04 - 15,330 * 0.06 = 93.40, or 0.934% of initial equity. A small change in either variable can erase it. Quoted APY is not a fixed return, and this calculation does not compensate for the 2.533x collateral exposure or tail losses.

Risks and controls

  • Liquidation and gap risk: debt interest, collateral declines or debt-asset appreciation can cross the liquidation boundary between observations. Use a stress-tested buffer rather than the maximum displayed borrowing power.
  • Correlation and depeg risk: assets described as correlated can separate. Stress each price independently and include redemption delays, wrapper exchange rates and slashing where relevant.
  • Rate and incentive risk: utilization can raise borrow rates, while supply rates and token rewards can fall. Model adverse rate paths and a zero-reward case.
  • Oracle and parameter risk: stale, capped or incorrect prices and governance changes can alter capacity or liquidation eligibility. Identify feeds, update rules, administrators and effective times.
  • Liquidity and unwind risk: supplied funds may be heavily borrowed and swaps may become thin. Map a manual deleveraging order and keep debt assets outside the loop for emergency repayment.
  • Smart-contract and integration risk: lending pools, tokens, wrappers, swaps, bridges and automation add separate failure and approval surfaces. Limit allowances and avoid dependencies that are not necessary.
  • Transaction risk: congestion, failed bundles, slippage and front-running can leave a partially built or partially unwound loop. Set explicit limits and verify state after every transaction.
  • Accounting risk: gross supply, debt, equity, rewards and TVL can be confused. Maintain an asset-liability ledger in one valuation unit and reconcile it on-chain.

Common misconceptions

  • “Repeated deposits compound wealth immediately.” They compound gross collateral and debt. Before market moves and costs, the algebra leaves initial equity unchanged.
  • “The theoretical leverage limit is the executable result.” The limit assumes identical rounds with no caps, fees, rounding or changing parameters and infinitely many transactions.
  • “Looping correlated assets is delta-neutral and cannot be liquidated.” Basis, depeg, wrapper, oracle and interest divergence remain, and liquidation follows protocol rules rather than a marketing label.
  • “Positive supply APY minus borrow APY guarantees profit.” Rates apply to different gross balances, can move independently, and rewards, gas, slippage and price exposure must be included.
  • “A protocol’s TVL increase represents new capital.” Recursive claims can increase gross measurements without the same increase in net equity; the result depends on the data provider’s accounting method.

Sources

Navigation

Search the wiki...