OverseasDeFi

OVERSEASDEFI / LEARN

DeFi Interest Rate Models: How Borrow and Earn Rates Are Set

How DeFi interest rate models set borrow and lending rates from utilization, what the kink does, and what BTC and ETH holders should check before borrowing.

OverseasDeFi·14 min read·Practical DeFi education

OverseasDeFi: know your positions, understand your risk.

Educational content only, not financial, investment, tax, or legal advice. DeFi and borrowing involve risk, including loss of capital. Income and appreciation are not guaranteed.

Abstract isometric lending rate curve

You open a DeFi lending app, see a borrow rate of 4%, and come back a week later to find it at 9%. Nobody called you. Nobody changed your contract. The rate moved because a formula did what it was built to do.

That formula is the interest rate model, and if you plan to borrow against Bitcoin or Ethereum instead of selling, it’s the thing that decides what your loan costs and what lenders earn on the other side. Short version: in most DeFi lending pools, the rate is a function of utilization, meaning how much of the pool’s money is currently borrowed. Low utilization, low rates. High utilization, rates climb, and past a set point (the “kink”) they climb fast.

This guide explains that in plain English, with real numbers from protocol documentation, so you can read a rate screen and know what could change under your feet.

Overseasdefi5steps.overseasdefi.comUnderstand DeFi Borrowing CostsLearn how variable rates work, what can change your borrowing costs, and how to assess the risks before putting strategies into practice.5 steps to earning daily Income

TL;DR

  • A DeFi interest rate model is a formula in a smart contract that sets borrow and lend rates automatically from pool utilization (the share of the pool’s supplied funds that is currently borrowed).
  • Most major models are “kinked”: rates rise gently up to a target utilization, then steeply above it, to pull in deposits and push borrowers to repay.
  • Lenders earn roughly the borrow rate times utilization, minus the protocol’s cut, so a pool with a 10% borrow rate does not pay depositors 10%.
  • If you borrow against BTC or ETH, a variable rate can change your carrying cost overnight, so stress-test the rate alongside your loan-to-value ratio.
  • Rates are published for a period and move constantly. Read the live protocol page, not a screenshot from last month.

Table of Contents

What an interest rate model actually is

Think of a pool of money like a rental car lot. If most cars sit on the lot, the company discounts to get them out. If almost every car is rented, prices go up because the last few cars are scarce. A DeFi lending pool does the same thing with money, except no manager sets the price. A smart contract (a program on a blockchain that runs exactly as written) does it.

Academic work describes the same design: lending protocols such as Aave and Compound use piecewise linear interest rate models that are functions of pool utilization. “Piecewise linear” just means the rate line is made of straight segments with a bend in it.

Utilization is the one number to know. JustLend’s documentation defines it as borrows divided by cash plus borrows minus reserves. If a pool holds 100 units of an asset and 80 are lent out, utilization is about 80%.

You may also see the phrase “interest rate mode” on lending apps. It usually refers to the choice or setting that decides how your borrow rate is calculated, such as a rate that floats with the pool versus one that is meant to be steadier. Which modes are available depends on the protocol and its version, so check the current options on the app you use before you assume a fixed rate exists.

What an interest rate model actually is

Why utilization drives the rate

The pool has two jobs that pull against each other. It needs borrowers, because their interest is what lenders earn. It also needs spare cash, because lenders must be able to withdraw.

The risk at the top end is real. JustLend’s documentation notes that as utilization approaches 100%, assets become scarce, so borrowing becomes impossible and suppliers may be unable to withdraw. LlamaRisk’s research makes the same point from the lender’s side: 100% utilization prevents withdrawals and leaves liquidators without inventory to work with.

So the model raises the price of borrowing as the pool empties. Chainflip’s documentation describes the idea simply: the curve raises borrowing costs when liquidity is scarce and rewards suppliers when demand rises.

The kink: where rates stop being gentle

Most models you’ll meet have two slopes. Below a target utilization, rates rise slowly. Above it, they rise steeply. The bend is the kink, or “optimal utilization.”

The Warwick paper on Compound explains that its model has a base rate (the rate at zero utilization) and a slope, with a kink at a threshold utilization typically 80%, above which a steeper slope applies. Governance voters, meaning holders of the protocol’s token, choose those parameters.

Here’s a worked example from Liquidium’s interest rate documentation. It uses example parameters of a 2% base rate, a 7% slope below the kink, a 300% slope above it and a 92% kink. These figures show the shape of the curve, not typical market rates:

Utilization Borrow rate (annual, example parameters) Where it sits
50% 5.8% Below the kink, gentle slope
92% 9% At the kink
98% 234% Past the kink, steep slope

Read that last row twice. Going from 92% to 98% utilization takes the borrow rate from 9% to 234% in that example. That isn’t a typo and it isn’t meant to be a normal price. It’s a penalty rate designed to make borrowers repay and lenders deposit, so the pool never runs dry.

Most of the time, a healthy pool sits near its target and you never see the steep part. But if you’re borrowing on a variable rate in a pool that’s pushed past the kink during a market panic, your cost can jump exactly when you least want a surprise. A paper on loanable-funds protocols adds that the kink may become a point where borrowers and lenders converge, which is why many pools hover close to it.

The kink: where rates stop being gentle

Why not just use a straight line?

The same paper describes three families of models: a simple linear rate, a non-linear one used by dYdX, and the kinked model used by Compound and Aave, all laid out with their formulas. Non-linear designs let rates rise at an increasing pace as utilization climbs, which gives stronger incentives to add liquidity and repay.

Protocols also tune curves to the asset. The Warwick paper finds that rules for riskier assets typically have higher slope parameters. Borrowing a stablecoin and borrowing a volatile token do not follow the same curve, and you shouldn’t assume the rate you saw on one applies to the other.

How the lender’s rate relates to your borrow rate

If you’re borrowing, you’re paying someone. If you later provide funds to a lending pool, you’re the someone being paid. The formula links the two.

Supply rate equals borrow rate, times utilization, times one minus the reserve factor. The reserve factor is the slice of interest the protocol keeps. Liquidium’s documentation gives an example: a 10% borrow rate at 80% utilization with a 10% reserve factor produces a 7.2% supply rate.

Walk through it: 10% times 0.80 is 8%. Take away 10% of that for reserves and you get 7.2%. Depositors earn less than borrowers pay for two reasons. Some of the pool is sitting idle, and the protocol takes a cut.

The cut varies. Suilend’s documentation describes a protocol spread (stated as 20% at the time of writing; check the current figure) and gives the same shape of formula. Others use a different structure altogether. Chainflip’s docs describe their own utilisation-based rate and fee setup, which may not follow the reserve-factor formula above, so check the current fee settings before comparing numbers across protocols.

Two lessons for a holder who wants income:

  1. A headline borrow rate tells you nothing about what a depositor earns.
  2. Different protocols take different cuts, so compare the formula and the fee, not just the advertised number.

If you want a realistic picture of what yields look like once costs are counted, our guide on realistic crypto yields for Bitcoin and Ethereum holders walks through it.

Variable versus fixed: the mortgage comparison

Most people already understand this tradeoff from home loans. A floating rate can start low and move with the market. A steadier rate costs more up front but is easier to budget. If you’ve compared fixed versus tracker rates on a mortgage, the logic transfers: the question is whether you can absorb a rate move.

The difference in DeFi is speed. A mortgage tracker moves when a central bank acts, a few times a year. A utilization-based rate can move block by block. Chainflip’s documentation, for instance, says interest is calculated every 10 blocks.

That’s why the practical question isn’t “what’s the rate today?” It’s “what happens to my plan if the rate doubles?”

What this means if you borrow against BTC or ETH

Borrowing against crypto is attractive if you don’t want to sell. You keep your Bitcoin or Ethereum, take a loan against it, and put that money to work. Our walkthrough on how to borrow against Ethereum without selling your ETH covers the mechanics. The interest rate model adds a second variable to a plan that many people only stress-test for price.

Here’s a simple way to think about it. Say you borrow against your holdings and deploy the loan somewhere that pays you a return. Your plan only works if what you earn stays above what you owe. A variable borrow rate can move up while your return stays flat, and the gap shrinks.

Before you open a position, run through these checks:

  • Find the pool’s utilization and its kink. Is it sitting at 55% or 88%? The closer to the kink, the less room before rates steepen.
  • Ask what happens at double the rate. If your plan breaks, the position is too thin.
  • Check the reserve factor and fees. They shape what lenders earn and, indirectly, how much liquidity a pool attracts.
  • Watch liquidity depth. A pool with a lot of idle funds is more forgiving than a small one that a few big borrowers can push past the kink.
  • Keep your loan-to-value (LTV) conservative. LTV is the loan as a percentage of your collateral’s value. Our guide to stress-testing your LTV shows how to do it.

Note that the rate is only half the story. Interest accrues to the loan balance, so a higher rate also nudges your LTV up over time even when prices don’t move.

Rate spikes and liquidations are linked

Interest rates and liquidations feel like separate topics. They aren’t. The Warwick paper finds that higher utilization raises a protocol’s net interest income but also raises systemic risks such as liquidations.

When prices fall hard, collateral gets liquidated (sold automatically to repay loans). A 2025 paper on risk-aware rates reports earlier research on contagion on Aave and Compound: liquidations push collateral prices down, which triggers more liquidations in a feedback loop, and pool defaults can expose lenders to losing part of their deposits.

The numbers from one stretch are sobering. The AgileRate paper reports that between 1 and 6 August 2024, Aave ETH V3 saw over $116M in liquidations after an ETH price drop. The authors estimate that a 15% higher over-collateralization ratio could have nearly halved liquidations for the affected users. They also argue that static curves struggled to hold the 0.8 target utilization they modelled when demand moved quickly.

The takeaway isn’t that borrowing is a bad idea. It’s that a calm-looking position can be tested by price and rates at the same time. Keep a buffer, and know in advance what you’d do if you got close to liquidation. Our piece on how crypto loan liquidation works and how to avoid it is a good next read, and so is how to reduce debt in DeFi safely.

Who sets the parameters, and can they change?

In many protocols, token holders vote on the curve. That means the kink, slopes and reserve factor can change after you’ve opened a position. Check the protocol’s governance forum or documentation for pending changes before you commit a large sum.

Researchers also debate whether fixed curves are good enough. The authors of AgileRate argue that static curves in Aave and Compound struggle to adapt to rapid market changes and propose an adaptive controller instead. LlamaRisk’s team puts the tradeoff this way: rate stability reduces responsiveness, which can leave markets at unsafe utilization levels for longer. In plain terms, nobody has a perfect model, and the design keeps evolving. Expect the rate screen you learn today to look a little different in a year.

Lending side: what to check before you deposit

If you’re providing funds to earn a rate, the same model works for you and against you. Higher utilization means higher yield, and it also means thinner withdrawal room. LlamaRisk explains that optimal utilization can be set from a withdrawal buffer: if peak withdrawals in volatile periods are 25% of supply, the protocol can operate safely up to about 75% utilization. The same worked example also applies a growth buffer so the market has room to grow. When peak withdrawals are very small, it sets optimal utilization at 85% instead of closer to 100%. That 85% works as a cap on the target, not a minimum, so a market with 25% peak withdrawals would still target about 75% (see a 15% growth buffer sets a floor of 85% for the full worked example). The final target reflects judgment about withdrawals and growth, not a single formula.

Before you deposit, look at utilization history, not only today’s number, and read our 15-minute pre-deposit check on liquidity pool risks. Tax matters too. In the US, crypto interest and rewards are generally treated as income when received, with gain or loss when sold, but treatment depends on your situation and jurisdiction. Our guide to US crypto rewards, income at receipt and gain or loss at sale explains the basics, and your CPA should confirm the details for your situation.

Putting it into a routine

A short weekly habit works better than a daily obsession:

  1. Check utilization and the borrow rate on each pool you use.
  2. Compare your loan’s rate to what your deployed funds are earning.
  3. Recheck your LTV against a price drop you’d consider painful.
  4. Note any governance proposals that change rate parameters.
  5. Write down your trigger: the rate or LTV level at which you repay or add collateral.

If you’re tracking several positions, a proper dashboard helps. See our guide to crypto portfolio tracking for DeFi borrowers.

Where we fit in

We built OverseasDeFi for people who hold Bitcoin or Ethereum and want it to do more than sit in a wallet, without day trading or selling. The Daily Yield System is our educational framework for working through Hold → Borrow → Deploy → Earn: whether to keep your crypto exposure, whether borrowing fits your situation, how deployed capital may earn, and whether the net result justifies the combined risks, including interest costs, liquidation and protocol risk. Members also get portfolio tools and an AI coach that can draw on portfolio context, health-factor checks and stress tests. These help you understand risk, but they can’t prevent liquidation or make decisions for you. Community and Accelerator offer different levels of support, alongside coaching, community and macro calls.

If you’d like a higher level of support, see the Accelerator. You make your own decisions; we focus on education, tools and guidance.

Overseasdefi

If you want to start small, OverseasDeFi offers a lower-commitment way to learn how rate models, LTV and risk fit together, alongside people asking the same questions.

FAQ

What is an interest rate model in DeFi?

It’s a formula in a smart contract that sets what borrowers pay and what lenders earn in a lending pool. The inputs are mostly utilization: the share of deposited funds currently borrowed. When utilization rises, rates rise, and the reverse.

What is the kink in a DeFi interest rate curve?

The kink is the utilization level where the curve bends from a gentle slope to a steep one. In the Compound markets studied by the Warwick paper, it was typically around 80%; other protocols and markets set it differently, such as the 92% kink in Liquidium’s example above. Past it, borrowing gets expensive quickly so that repayments and deposits restore liquidity.

Why do lenders earn less than borrowers pay?

Only the borrowed share of the pool generates interest, and the protocol keeps a portion as reserves or a fee. In Liquidium’s example, a 10% borrow rate at 80% utilization becomes a 7.2% supply rate.

Can a DeFi borrow rate change after I take a loan?

On a variable setup, yes, and it can change often as utilization moves. Governance votes can also change the curve itself. Build your plan so that it still works if the rate is meaningfully higher than when you started.

Is a higher DeFi yield always better?

Not automatically. A high rate can mean high utilization, which can leave less room to withdraw. Read the pool’s liquidity and risks alongside the number.

OverseasdefiDiscuss DeFi Rates and RiskReach out with questions about borrowing against crypto, rate changes, and understanding the risks before putting a strategy into practice.OOverseasdefiEmail usRoman.rida@overseasdefi.com+1 310 569 7547

Sources