APR vs APY in DeFi: compare annual rates and actual returns

In DeFi yield quotes, APR expresses an annual rate without compounding; APY includes compounding under the stated assumptions. Neither is the percentage you necessarily earn this month. Before comparing two offers, check the asset, period, reinvestment rules and costs behind each number.

You can do this comparison in a worksheet without connecting a wallet or moving funds. Start with the provider’s rate explanation and a dated quote. The examples below are hypothetical calculations, not current yields or forecasts.

Establish what each percentage measures

The Aave glossary distinguishes APY from APR by whether compounding is included. Compounding means earlier earnings also participate in later earnings. The distinction changes the arithmetic even when the underlying nominal rate is identical.

Label	What the figure includes	What you must still check
APR	Annual rate without compounding	Whether earnings can be reinvested, and when
APY	Annual effect with assumed compounding	Which assumptions the quote uses
Period return	Change over a stated interval	Flows, prices, rewards and costs

These definitions describe DeFi yield comparisons. Do not assume every lending disclosure or banking APR uses exactly the same fee conventions.

Also distinguish a supply rate from a borrowing rate. One describes compensation on supplied assets; the other describes a cost on debt. An attractive supply percentage does not resolve the risk of a loan. If you already borrow on Aave, use the separate health-factor check for that question.

Calculate the compounding difference once

Take 1,000 units of a hypothetical token earning a fixed 12% nominal APR for one year. Assume no fees, price changes, extra deposits or withdrawals.

Without reinvesting earnings, the model produces 1,000 × 0.12 = 120 token units. The ending amount, including the separately retained earnings, is 1,120 units.

Now assume the same nominal rate is credited in 12 equal monthly periods and every month’s earnings immediately earn the same rate. Each period’s rate is 0.12 / 12 = 0.01, or 1%. The annual result is:

APY = (1 + APR / n)^n - 1

Here n is 12, so (1 + 0.12 / 12)^12 - 1 = 0.12682503, approximately 12.6825% APY. The ending amount is approximately 1,126.83 token units.

The extra 6.83 units come from the assumed reinvestment. They are not a bonus inherent in the letters APY. The Coinbase explanation also separates simple APR from APY’s compounding effect.

Use this formula only when the nominal rate and compounding schedule are established. A reward paid in another token and left uninvested does not automatically earn the base rate. If a provider already quotes APY, do not put that APY into the APR formula and compound it again.

Convert an annual quote to the period you need

An annualized rate expresses a year-equivalent measure. The annualized label alone does not establish a one-year commitment by the provider. A 12% annual quote is not a 12% monthly return.

For simple interest at a constant 12% APR over 30 days, using a 365-day convention:

1,000 × 0.12 × 30 / 365 = 9.8630 token units

For a hypothetical constant 12% APY, the equivalent effective return over 30 days under a uniform growth model is:

(1 + 0.12)^(30 / 365) - 1 = 0.00935820

That is approximately 0.9358%, or 9.36 token units on 1,000 units. A calendar month’s length and a provider’s accrual rules may differ. This conversion is an illustration of a fixed annual model, not an exact reconstruction of a protocol balance.

The two 12% figures are different inputs. The APY already incorporates compounding across the year, whereas the APR does not. To derive an equal monthly rate from APY, use (1 + APY)^(1 / 12) - 1, not APY divided by 12.

Keep a changing rate out of a fixed-rate forecast

Aave’s supply guide explains that supply rates update as liquidity conditions change and depend on utilization and governance parameters. A rate you read today does not establish the rates that applied throughout your holding period.

Consider another hypothetical simple-interest model: 1,000 token units earn 12% APR for 15 days, then 6% APR for 15 days. Keep principal unchanged and retain earnings separately. Interest is:

1,000 × (0.12 × 15 / 365 + 0.06 × 15 / 365) = 7.3973 units

Using the first quote for all 30 days would instead give 9.8630 units. The difference comes from the changed input. Neither calculation is an Aave accrual simulation. Actual protocol accounting needs its own historical records and rules.

For an existing position, record actual accrued amounts where available. Do not manufacture a rate history from one current screenshot. An interval with missing data remains incomplete.

Complete one comparison worksheet

Create one row per position or quote, then fill these fields before ranking percentages:

1. Identity and snapshot. Provider, network, market, token contract, quote time and whether the rate is current or historical.
2. Rate basis. APR or APY, supply or borrow, base rate or combined incentives, and fixed or variable.
3. Compounding. Frequency and whether reinvestment is automatic, conditional or a separate action. Leave it unknown if documentation is unclear.
4. Common period. Use the same start and end dates and state the day-count convention for any model calculation.
5. Asset and valuation. Record token units separately from their value in one chosen currency. Give reward tokens their own rows and dated prices.
6. Costs and flows. Note fees already included in the quote, additional costs, deposits and withdrawals. Do not subtract an included fee twice or call a new deposit earnings.

A token-denominated gain can accompany a fall in dollar value. A pool return can also differ from simply holding the original assets; the impermanent-loss comparison examines that benchmark separately. Normalizing rate labels does not make the positions equivalent in risk.

A portfolio balance is a snapshot of value, not proof of earnings. Finish the worksheet with what is known, the assumptions used and any missing fields. You then have a rate comparison you can explain, rather than a precise-looking prediction made from one percentage.