Why an Annualised Funding Rate Is Not a Return

A funding rate is quoted per settlement interval — typically per eight hours. Annualising it means multiplying by the number of settlements in a year: a rate of 0.01% per eight hours, settled three times daily, annualises to 10.95% simple, or about 11.57% if compounded.

That conversion is arithmetically correct and practically misleading, because it carries four assumptions that are rarely true at once: that the rate holds for a year, that execution is free, that the sign never inverts, and that any amount of capital can earn it. A realised return is what survives all four.

This matters beyond pedantry. Annualised figures are how this entire category advertises itself, and the distance between the headline and the outcome is where most of its credibility problems live.

The Arithmetic

Simple Annualisation

Multiply the per-interval rate by the number of intervals in a year.

Rate × settlements per day × 365

At the common baseline of 0.01% per eight hours: 0.01% × 3 × 365 = 10.95%. This is an APR — an annual percentage rate, ignoring any reinvestment.

Compounding, and Why the Two Numbers Differ

If each payment is reinvested, the result compounds across 1,095 intervals:

(1 + rate)^(settlements per year) − 1

At the same 0.01%: (1.0001)^1095 − 1 ≈ 11.57%. This is an APY — an annual percentage yield, including compounding.

The gap between 10.95% and 11.57% is modest at the baseline and grows sharply at higher rates, which is where annualisation starts to mislead badly:

A rate near the upper boundary of a venue's funding formula annualises to a number that looks extraordinary. It is also the rate least likely to persist — venue formulas are built to pull the premium back toward baseline, which is exactly what a sustained extreme rate incentivises participants to do. The largest annualised figures are almost always the most transient.

This is also the first place to check any quoted yield: an APY and an APR calculated from the same underlying rate are different numbers, and nothing obliges a crypto product to tell you which it used.

Four Assumptions Hidden in the Headline

One: It Annualises a Moment

Annualisation takes a rate observed now and projects it across a year. Funding rates change every settlement, cluster at the baseline for long stretches, spike during positioning extremes, and revert.

Projecting a single print forward is not a forecast; it is a unit conversion. "This rate, if it held for a year" is a statement about arithmetic, not about the market — and in a year like 2026, in which bitcoin perpetual funding spent more than thirty consecutive days below neutral, any single positive print projected across the year would have been badly wrong. We covered that episode in what a negative funding rate means.

Two: It Is Gross of Costs

The annualised figure describes the funding payment. It says nothing about what it costs to be in a position to receive it.

A delta-neutral position requires two legs. Opening means crossing two spreads and paying fees on both; closing means doing it again. At a baseline funding rate, the income accrues at roughly 0.01% per eight hours — so a round trip costing a few basis points consumes several days of accrual before the position is net positive. Positions rotated frequently can spend much of their gross income on execution. There is also collateral held against the short leg that is not itself earning.

Gross funding and net return are different numbers, and only one of them is usually advertised.

Three: It Assumes the Sign Never Flips

Annualising a positive rate assumes positive funding for a year. Inversions are not edge cases; they are a recurring feature, and the arithmetic is symmetrical — a −0.01% rate costs exactly what +0.01% paid.

A realistic annual figure is the net of positive and negative intervals across the whole period, not an extrapolation from the favourable ones.

Four: It Assumes Capacity

A funding rate is available to the capital that can actually be deployed into it at that rate.

Establishing a large position means moving through the order book on both legs, and the spread available to a small position is not necessarily available to a large one. Beyond a certain size, deploying more capital compresses the very spread being captured. Annualised figures implicitly assume unlimited capacity at the quoted rate, which is never true and matters more as size grows.

Time-Weighted, Not Point-in-Time

The defensible way to express what a funding strategy earned is a time-weighted average of realised funding across the period, net of costs — not an annualisation of a selected rate.

The difference is not cosmetic. A period containing a few days of extreme funding and many weeks at baseline has an average far closer to baseline than to the peak. Any figure derived from the favourable part of a distribution describes the distribution's tail, not the experience of holding a position through it.

When a figure is quoted, the questions are: over what period, net of what, and including the negative intervals?

In Regulated Savings, the Method Is Prescribed

Here is a comparison worth sitting with.

In US consumer deposit accounts, the annualisation method is not a matter of choice. The Truth in Savings Act and its implementing Regulation DD define annual percentage yield as a rate reflecting total interest paid, based on the interest rate and the frequency of compounding over a 365-day period, calculated according to rules set out in an appendix to the regulation — with the Federal Reserve's compliance guide to Regulation DD summarising the framework. Institutions must disclose both the interest rate and the APY, and must give advance notice of changes that would reduce it.

The point of a prescribed formula is comparability. When every institution computes the number the same way, two advertised figures can be compared directly.

Crypto yield products have no equivalent. Different products annualise differently, compound differently or not at all, quote gross or net on their own terms, and select periods at their discretion. Two advertised figures from two providers are frequently not measuring the same thing, and no rule requires them to.

That is not an argument that crypto yield is illegitimate. It is an argument for reading the method rather than the number — and for being sceptical of any provider that does not publish one.

What a Meaningful Figure Would Look Like

A quoted figure is interpretable when it states:

• The period it covers, and whether that period was representative.
• Simple or compounded, and at what frequency.
• Gross or net, and net of precisely which costs.
• Whether negative intervals are included or excluded.
• At what size the figure was achieved, and whether it scales.

A number without those is a marketing figure. A number with them can be examined — which is a different kind of claim.

Questions to Ask About Any Quoted Yield

1. Is this APR or APY, and what compounding frequency is assumed?
2. Is it an extrapolation of a current rate, or a realised result over a stated period?
3. What costs are deducted, and which are not?
4. Does the period include drawdowns and negative funding, or only favourable conditions?
5. Does it assume capacity at that rate for the size being considered?
6. Who calculated it, and against what methodology?

The same discipline applies to the sources of yield themselves, which we set out in where stablecoin yield comes from.

Frequently Asked Questions

How do you annualise a funding rate?

Multiply the per-interval rate by the number of settlements in a year — for an eight-hour interval, three per day across 365 days, or 1,095 intervals. For a compounded figure, raise one plus the rate to the power of the number of intervals and subtract one.

What is the difference between APR and APY?

APR is a simple annual rate that ignores reinvestment. APY includes compounding. At a 0.01% eight-hour funding rate the two are approximately 10.95% and 11.57%; the gap widens substantially at higher rates.

Is an annualised funding rate a forecast?

No. It is a unit conversion stating what a current rate would produce if it persisted for a year. Funding rates change every settlement and can turn negative, so an annualised figure describes arithmetic rather than expectation.

Why is the realised return lower than the quoted rate?

Because the quoted rate is typically gross and point-in-time. Execution costs on two legs, collateral that is not itself earning, periods of negative funding, and capacity constraints at larger sizes all sit between the headline and the outcome.

Why are crypto yield figures hard to compare?

Because no prescribed calculation method exists. Regulated deposit products must compute APY according to a defined formula so that figures are comparable. Crypto products annualise, compound, select periods and define costs at their own discretion, so two advertised numbers may not be measuring the same thing.

The Number Is the Smallest Part of the Answer

Annualisation is a useful tool for putting a per-interval rate into familiar units. It becomes misleading the moment it is read as a prediction of what a year of holding would produce.

The four assumptions behind every annualised figure — persistence, zero cost, constant sign, unlimited capacity — are each individually reasonable as simplifications and collectively almost never true. Which is why the method matters more than the number, and why any provider unwilling to publish the method has told you something about the number.

The mechanics underneath are set out in our breakdown of how the funding fee works, and the full strategy construction in our guide to funding rate arbitrage.

This article is published for informational purposes and does not constitute investment advice or an offer to transact. All figures shown are illustrative arithmetic conversions of hypothetical funding rates, not indications, projections or guarantees of return. Funding rates are variable and can turn negative. Capital is at risk.

This article is for information only and is not investment advice. Past performance does not guarantee future results.