# APY Is Not One Number

**Published by:** [yearn](https://blog.yearn.fi/)
**Published on:** 2026-08-18
**Categories:** ethereum, vaults, defi, yield, yearn, uniswap, morpho, aave, pendle
**URL:** https://blog.yearn.fi/apy-is-not-one-number

## Content

Every DeFi dashboard puts a percentage beside a vault, pool, or market. That makes yield numbers look easy to compare. A 5% APY should beat a 4% APY, right? Not necessarily! APY numbers are often more complicated than a single summary number can explain. One protocol may annualize a live interest rate. Another may multiply yesterday's trading fees by 365 to annualize it. A third may measure a vault share's growth over the last month and extrapolate it. All 3 methods produce a percentage, but they are not measuring the same thing. Two protocols can display the same APY while measuring different things, and there is no clear correct answer. Although it is never advertised or taught anywhere, users need to bring their own awareness and knowledge to the table to evaluate their yield options. This article examines the APY calculation paths for 11 protocols: Yearn v3, Compound v2, Aave v3, Uniswap v2/v3/v4, Curve, Lido, Sky, Morpho, Euler, Pendle, and GMX. The differences lie in the formulas and in the assumptions behind them. The short version is found in this summary table: Protocol Yield source Time basis Calculation style Yearn v3 Vault share-price growth Weekly, monthly, and inception lookbacks Compounded annualization Compound v2 cToken lending interest Current per-block rate Daily compounding Aave v3 Reserve lending interest Current per-second rate Per-second compounding Uniswap v2/v3/v4 LP trading fees Trailing 24-hour volume Linear annualization (no compounding) Curve Base pool fees plus CRV and incentive emissions Daily/weekly base data; current rewards API-reported base APY plus linear rewards math Lido Staking rewards Latest rebase or moving-average window Linear annualization Sky Governance-set savings rate plus separate rewards Current parameter projected over one year Per-second compounding for savings; linear rewards math Morpho Supply yield derived from borrow rate, utilization, and fee Current IRM and utilization Continuous compounding in the SDK Euler Onchain lending rate plus intrinsic yield Current lending rate; backend intrinsic-yield data Per-second onchain APY plus combined intrinsic APY Pendle PT market-implied yield Spot or TWAP price plus time to maturity Term-based annualization GMX Trading-fee yield, incentives, and staking pass-through Selectable trailing fee period; per-program rewards Backend fee APY plus linear incentive APR What the APY measures, and what hidden expectations exist The summary table explains the yield sources. The risk sits in what the expectations are to maintain the current yield number. Different yield sources have different measurement approaches and carry different volatility. Comparing different APYs blindly is like comparing a mortgage rate with a stock's trailing dividend yield, just because both are percentages. Borrow-side yield depends on demand and utilization. Trading-fee yield depends on volume, fee tiers, and the behavior of the pool's assets. Staking yield depends on validator performance and the staking system. Governance-set yield can change when governance changes a parameter. Incentives end when the program ends. Two protocols can therefore show the same APY while exposing a depositor to very different expectations about maintaining the current yield. A change in market borrowing conditions can dramatically impact one source of yield, but not others. Some yield sources can be much more stable than other more volatile sources. The number alone does not provide information about what conditions must be kept stable for the return to stay constant. Some APYs look forward. Others look backward. APY numbers can be based on historical results, current rates, or forward projections. Different protocols take different approaches. Some protocols use the current rate. Aave v3, Compound v2, Morpho, and Euler display an annualized version of the rate available now. The calculation assumes that today's rate continues for a year. If there is any change in utilization, then the displayed number changes with it. This rate is obviously quite variable, but because lending markets make it easy to know the exact value of the current rate, it makes sense to use the latest information available. Other protocols use a period of historic rate data. Yearn, Uniswap, Lido, Curve's base APY, and GMX's fee component use realized activity from a lookback window. Yearn compares weekly, and monthly windows. Uniswap's fee estimate uses roughly the last 24 hours of volume. Lido annualizes the latest rebase. Curve's base APY uses daily or weekly data. GMX lets the interface select a trailing period for its fee data. Curve and GMX calculate their rewards or incentive components separately from current emissions or program data. Finally, some protocols use their own approach. Sky extrapolates its current governance-set savings rate forward over a year. Pendle uses a market-implied rate derived from PT pricing and time to maturity, either from a spot price or a TWAP. Neither is a lookback measurement of realized depositor returns. A short time window reacts quickly, but it looks very volatile. A single day of unusual volume can make a pool look exceptional. A longer window is steadier, but it can hide a recent change in conditions. These approaches answer different questions: What rate is available if current conditions persist? What return did depositors actually receive recently? What rate is implied by current market pricing? A displayed APY is more useful once you know which question it answers. Rewards and bundled headlines Some headline APYs include reward tokens. To include these rewards into the APY number, protocols usually annualize current token emissions and convert them into dollars at the current value: reward tokens per year × token price ÷ deposited capital The result is still an estimate. If the reward token's price rises, the displayed dollar APY rises even when emissions have not changed. If the token price falls, the depositor receives less dollar value. When the incentive program ends, that part of the APY disappears entirely. Curve and GMX make this distinction visible by separating base yield from reward or incentive components. Sky can also show a governance-set savings rate alongside separate rewards math. These components should not be read as one homogeneous return. Ask: what token do I receive, and what happens when its price changes? Not every APY comes from one input. Curve can combine a base APY with a rewards estimate. GMX can combine trailing fee APY, incentive APR, and a staking component. Euler can show the lending return from its onchain lens alongside an intrinsic APY for the underlying asset. Aave v3 displays base reserve rates and incentive APR separately. A bundled headline can be useful, but its components may use different data sources, windows, compounding rules, token denominations, and expiration dates. Split it apart and consider the yield components separately. Ask which part is earned by the underlying strategy and which part is paid to subsidize deposits. Ask whether each part is measured or projected. Native interest, trading fees, and staking rewards are not risk-free. But an emissions-based APY adds another variable to the calculation. The reward token's market price becomes part of the return path. APR and APY use different compounding assumptions APR is a simple annualized rate. APY assumes that returns are reinvested and compounded. The difference is small at low rates and grows quickly at high rates. For example, 10% APR compounded daily is a 10.5% APY. Protocols also compound at different intervals. Compound v2 uses a daily convention for its frontend APY. Aave v3, Euler, and Sky use per-second compounding calculations. Morpho uses continuous compounding in its SDK. Uniswap's fee estimate is linear annualization, so it uses APR math even when the surrounding interface uses the word yield. The same underlying rate can produce different headlines depending on the convention. A number labeled APY does not tell you whether the protocol compounds daily, per second, continuously, or not at all. The short version Before comparing APYs, ask: What is the source of yield: interest, fees, share-price growth, staking rewards, emissions, a governance-set savings rate, or market-implied yield? Is the number from a live snapshot, a trailing average, or a forward projection? What lookback window feeds the number? Is the return paid in the deposited asset or in a reward token? Is the headline one return or several components added together? Does the calculation include compounding, and how often? An APY number is not the yield itself. It is a model of yield with assumptions about time, compounding, prices, and what tokens are counted. Once you can name and understand those assumptions, you can decide whether two yield numbers from different protocols are actually comparable.

## Publication Information

- [yearn](https://blog.yearn.fi/): Publication homepage
- [All Posts](https://blog.yearn.fi/): More posts from this publication
- [RSS Feed](https://api.paragraph.com/blogs/rss/@yearn): Subscribe to updates
- [Twitter](https://twitter.com/yearnfi): Follow on Twitter