Quant Memo
Core

DeFi Rate Arbitrage and Looping

If a lending protocol pays more to supply an asset than it costs to borrow a related one against it, a trader can repeat the deposit-borrow-redeposit cycle to multiply that spread many times over — at the cost of multiplying liquidation risk by the same factor.

Prerequisites: Proof of Stake and Staking Yields, Margin Calls and Forced Liquidation

DeFi lending protocols set supply and borrow rates for each asset independently, driven by utilization — how much of the deposited pool is currently borrowed. It is common for a protocol to pay, say, 3% to suppliers of a liquid-staking token while charging only 2% to borrow the stablecoin used to acquire more of it. Whenever a supply rate on one asset exceeds the borrow rate on a correlated asset used to fund it, a trader can capture that spread — and can do it more than once by looping: deposit collateral, borrow against it, use the borrowed funds to buy more of the collateral asset, deposit that too, and repeat.

Looping repeats a deposit-borrow-redeposit cycle to multiply a rate spread across several times the trader's original capital, but every loop also multiplies leverage — so the same mechanism that scales up the yield scales up liquidation risk by the same factor.

How the loop compounds

1. deposit collateral 2. borrow 3. buy more collateral 4. redeposit
Each pass around the loop adds another layer of borrowed exposure on top of the same underlying spread — leverage and yield scale together.

Protocols cap this with a maximum loan-to-value ratio, which limits how many loops are actually possible: each pass through the loop only unlocks a fraction of the previous deposit as new borrowing capacity, so the position converges to a maximum leverage multiple rather than growing without bound.

Worked example

Suppose a liquid-staking token yields 3.2% and can be posted as collateral to borrow a stablecoin at 1.8%, with a maximum loan-to-value of 75%. Starting with $10,000 of the staking token:

  1. Loop 1: deposit $10,000, borrow 10{,}000 \times 0.75 = \7{,}500$, buy more staking token with it.
  2. Loop 2: deposit that $7,500, borrow 7{,}500 \times 0.75 = \5{,}625$.
  3. Loop 3: deposit $5,625, borrow 5{,}625 \times 0.75 = \4{,}219$.

Continuing this geometric series to its limit, total collateral held converges to 10{,}000 / (1 - 0.75) = \40{,}000, funded by \30,000 of borrowing — a 4x leverage multiple on the original $10,000. Net yield: 40{,}000 \times 3.2\% - 30{,}000 \times 1.8\% = 1{,}280 - 540 = \740peryear,roughly7.4 per year, roughly 7.4% on the original \10,000, more than double the raw 3.2% spot yield.

Leverage in a loop is a fixed multiple applied to a small spread, which means it is also a fixed multiple applied to any drop in the collateral asset's price. If the staking token falls just 5% while the borrowed stablecoin holds its value, the 4x-leveraged position loses roughly 20% of the trader's original equity — and if that erases the loan-to-value buffer, the position gets liquidated at a discount, wiping out several years of the captured spread in one event.

Related concepts

Practice in interviews

Further reading

  • Aave Protocol Documentation, Interest Rate Model
ShareTwitterLinkedIn