Funding Rate Benchmark: Hyperliquid vs CEX (Live, Open Data)

This article contains affiliate links. Yieldo may earn a commission at no extra cost to you.

Updated 02 October 2026

The dataset summarizes observed API fundingPeriodic payments between buyers and sellers of perpetual futures. Buyers pay when the rate is positive; sellers pay when it is negative. This mechanism keeps the contract price closer to the underlying asset price. quotes after interval normalization, not settled payments or realized strategy returns. Hyperliquid, Aster and edgeX appear as separate venue groups when eligible data exists. The CEX group uses the median of available exchange values; membership follows active providers and data validity, not a permanent count of eight venues. Missing or rejected observations are not zero funding.

TL;DR — The Number, the Method, the Date

• The dataset summarizes observed API funding quotes after interval normalization, not settled payments or realized strategy returns. Hyperliquid, Aster and edgeX appear as separate venue groups when eligible data exists. The CEX group uses the median of available exchange values; membership follows active providers and data validity, not a permanent count of eight venues. Missing or rejected observations are not zero funding.
• The method: inside a single exchange, the benchmark averages that exchange's own observations over the window. Across the CEX group, it takes the median of the per-exchange averages — not the mean — because one thinly-traded exchange with a stale or outlier print shouldn't move the whole group's number. Every rate is annualized first: rate × (24 / interval_hours) × 365 × 100.
• The date: this benchmark's history starts 2026-07-05 (coverage_from in the feed). That's a hard floor — we don't claim, estimate, or backtest anything earlier. Aster is a separate, later case: its funding history in this benchmark only begins 2026-08-04, so treat any Aster figure as a fresh data point, not a month of history.
• The chart below can legitimately be sparse. A venue's line only appears once it has at least two days of history; Aster's line may not be there yet on the day you're reading this, and that's expected behavior, not a bug.
• Where to see the live numbers: the chart and table further down this page, or pull the raw feeds directly — JSON and CSV, both free, both CC BY 4.0 licensed, both refreshed every five minutes.

Live Funding Rate Benchmark: Hyperliquid vs CEX Median

The dataset summarizes observed API funding quotes after interval normalization, not settled payments or realized strategy returns. Hyperliquid, Aster and edgeX appear as separate venue groups when eligible data exists. The CEX group uses the median of available exchange values; membership follows active providers and data validity, not a permanent count of eight venues. Missing or rejected observations are not zero funding.

The historical chart averages observations within each exchange’s available instrument set before combining CEX values by median. The current table uses selected coins. These are different slices: coverage changes can move the chart, so it is not a matched-contract execution study. Use per-coin rows for a like-for-like comparison.

7-day and 30-day annualized windows

Underneath the daily chart, the benchmark also tracks two rolling windows for every venue-coin pair: a 7-day average and a 30-day average, both annualized the same way as the chart. The 7-day window reacts faster to a recent shift in positioning; the 30-day window smooths out noise from any single volatile week. Neither window is a prediction — both are backward-looking averages of what actually printed, recalculated as new data lands. A snapshot of the current, unaveraged rate sits alongside both windows in the underlying feed, so you can see the instant reading and the smoothed trend side by side.

Reading the chart (and what an empty chart means early on)

A venue's line only draws once its group has at least two distinct days of history in the lookback window; a group with a single day of data is dropped from that day's render rather than shown as a flat or fabricated line. In practice that means Hyperliquid and the CEX median — both backed by a full month of history since coverage_from — should render normally, while Aster's line, tracked separately since 2026-08-04, will fill in day by day as its own history accumulates and may not be visible yet as of 02 October 2026, depending on exactly when you're reading this. If every tracked group happens to have fewer than two days of data at once, the widget shows an explicit message instead of a blank chart or a misleading zero line — that's the benchmark being honest about what it does and doesn't have yet, not a rendering failure.

Methodology: How We Normalize and Compare Funding Rates

Everything above depends on getting three things right: converting each rate to a common time basis, choosing the right way to combine multiple exchanges into one number, and being upfront about which venues get averaged versus grouped. This section documents exactly how FundingBenchmarkService, the code that produces this benchmark, does each one — reproducible on purpose, so you can verify it against the raw feed rather than take our word for it.

Annualization formula — why raw 1h and 8h rates aren't comparable

Perpetual futuresA contract that tracks changes in an asset price without a fixed expiry date. Trading requires collateral; a position may be forcibly closed if collateral is insufficient. Periodic payments between the two sides are called funding. exchanges don't share a funding clock. Hyperliquid settles funding once every hour, on every pair, continuously. Most of the centralized exchanges in this benchmark's CEX group settle on an 8-hour base interval instead (a handful apply a shorter interval on select volatile pairs, but 8 hours is the standard default). A raw 0.01% print on an hourly venue and the same 0.01% print on an 8-hour venue are not the same amount of money over a year — the hourly venue charges or pays that rate eight times more often. Lining the two raw numbers up side by side and calling the bigger one "the better rate" is the single most common mistake in funding rate comparisons, and it's exactly what this benchmark exists to prevent.

The fix is one formula, applied consistently to every observation before anything else happens to it:

annualized_rate (%) = rate × (24 / interval_hours) × 365 × 100

Here rate is the exchange's raw fractional rate at each funding event, and interval_hours is that specific venue's settlement interval — 1 for Hyperliquid, 8 as the CEX default (with per-pair exceptions on some exchanges). Every figure this benchmark shows, in the chart, in the current-snapshot table, and in the downloadable feed, is already this formula's output. Nothing you see was compared before being converted to a common annual basis.

Why the CEX group uses the median, not the average

The dataset summarizes observed API funding quotes after interval normalization, not settled payments or realized strategy returns. Hyperliquid, Aster and edgeX appear as separate venue groups when eligible data exists. The CEX group uses the median of available exchange values; membership follows active providers and data validity, not a permanent count of eight venues. Missing or rejected observations are not zero funding.

This benchmark uses the median of the per-exchange averages instead, which is why the field in the public feed is literally named cex_median, not cex_average. A median picks the middle value once the exchanges are sorted, so a single outlier exchange has to drag several of its neighbors along with it before it can move the group's reported number at all. For a single-venue group like Hyperliquid or Aster, where there's exactly one exchange to combine, the median of one value is trivially that value — the formula doesn't change behavior there, it just stays consistent across the codebase.

This distinction is not academic: an earlier version of the feed-generation code combined already-averaged per-exchange figures back together using a mean of means instead of a true cross-exchange median, which meant the field named cex_median was not, for a period, actually a median. That bug was caught in review and fixed. We're mentioning it here on purpose — a benchmark that gets its own headline statistic wrong isn't trustworthy just because it claims a rigorous method in its documentation, and a data source that's willing to describe a defect it found and fixed in its own methodology is more verifiable than one that only ever describes success.

Single-venue groups (Hyperliquid, Aster): mean of own observations

Within any single exchange — including single-venue groups like Hyperliquid and Aster, which each get their own line rather than being folded into "CEX" — the benchmark computes a straightforward average of that exchange's own annualized observations over the window in question. There's no cross-venue aggregation happening at this step; it's the same exchange's own funding prints, averaged over time. The median-versus-mean question only applies once you're combining different exchanges into a single group number, which is why it's specific to the CEX group and doesn't change how Hyperliquid's or Aster's own line is calculated.

Data Coverage and Limits

A benchmark is only as trustworthy as its stated limits. Here's exactly what this one covers, what it doesn't, and where the boundary sits.

Coverage starts July 5, 2026 — no backtests deeper than that

Yieldo began recording funding rate history for this benchmark on 2026-07-05. That date is a hard floor, not a rounding choice: the underlying database table simply has no rows before it, and the public feed carries this exact date in its coverage_from field on every request so downstream consumers don't have to guess. Nothing in this article, the chart, or the feed represents a backtest, an estimate, or a reconstruction of funding rates before that date. If you need funding history from earlier than mid-2026, this dataset is not the source for it — say so plainly rather than stretch the claim.

Aster: recorded only from August 2026, treated separately

Aster is included in this benchmark as its own venue group, but its history in Yieldo's tracker only begins on 2026-08-04 — a full month after the benchmark's general coverage_from date for Hyperliquid and the CEX group. In practice this means any 7-day or 30-day average shown for Aster early on is built from however many days have actually accumulated since 2026-08-04, not a full window — the underlying calculation doesn't pad missing days with assumptions, it simply averages what's there. Whenever this article discusses Aster numerically, that shorter, more recent start date applies; it should never be read as carrying the same depth of history as Hyperliquid or the CEX median.

What this benchmark does not claim

The dataset summarizes observed API funding quotes after interval normalization, not settled payments or realized strategy returns. Hyperliquid, Aster and edgeX appear as separate venue groups when eligible data exists. The CEX group uses the median of available exchange values; membership follows active providers and data validity, not a permanent count of eight venues. Missing or rejected observations are not zero funding.

Hyperliquid vs CEX: What the Data Shows Right Now

The historical chart above shows the trend; the table below shows a snapshot — where each tracked venue stands against the CEX median for the same coins right now, accurate as of 02 October 2026, plus the 7-day and 30-day windows for context.

Perp DEX vs CEX funding — current snapshot

The Δ column in the table above is the gap, in annualized percentage points, between the median across perp-DEX venues and the CEX median for the same coin — not any single venue's own rate, so it can differ from what you'd get comparing one DEX's column straight against the CEX median column (the formula holds unchanged when only one perp-DEX venue is live, since the median of a single value is that value) — the same underlying comparison this benchmark's chart tracks over time, just read at a single moment. Coverage here follows config('coins.popular'), Yieldo's tracked list of popular coins (BTC, ETH, USDT, USDC, SOL, BNB, XRP, TON, ADA, DOGE, HYPE), not every instrument every exchange lists. For a full pair list and history specific to one venue, see the Hyperliquid funding page or the Aster funding page; for BTC and ETH funding broken out across every exchange Yieldo tracks, see the dedicated BTC funding page and ETH funding page.

Download the Dataset: JSON and CSV Feeds

The chart and table above are built from the same public feed anyone can pull directly — the same code, the same numbers, no separate "public" dataset that quietly differs from what's rendered on this page.

Fields and update frequency

The JSON feed lives at https://yieldo.me/data/funding-benchmark.json and returns, per entry in its venues array, the venue name, the coin symbol, current_annualized (the latest unaveraged reading), avg_7d, and avg_30d — all in the unit the payload declares itself: annualized_percent. Each entry also carries a boolean extreme: true when any of its three annualized values is above 50% in absolute terms, meaning that figure is an extreme, not an annual yield; the numbers themselves are published unchanged. The top level of the payload also carries coverage_from, a methodology link back to this article, and an updated_at timestamp. The CSV twin at https://yieldo.me/data/funding-benchmark.csv carries the same fields as plain columns — venue, coin, current_annualized_percent, avg_7d_percent, avg_30d_percent, coverage_from, extreme — with the attribution written into a leading comment line rather than a separate metadata file, so it survives even if someone strips the header row. Both endpoints are served with Cache-Control: public, max-age=300, meaning they're safe to poll on a five-minute cadence without hammering the origin — no API key required.

License and attribution (CC BY 4.0)

Both feeds are published under CC BY 4.0 (Creative Commons Attribution 4.0) — free to use, redistribute, and build on, commercial use included, with attribution to Yieldo required. Every JSON response and every CSV file carries this license, plus a machine-readable attribution block, directly in the payload itself: source, license, license_url, and a ready-made cite_as string, so the attribution travels with the data even if it's copied into another tool without the surrounding page.

How to cite this dataset

The feed's own cite_as field spells out the expected citation directly: "Source: Yieldo (yieldo.me) — real-time exchange data aggregated from official APIs." When citing a specific reading, add the date you pulled it and, ideally, a link back to this methodology page — the full data catalog, including this dataset alongside Yieldo's other open feeds, is indexed at Yieldo's Free Data API catalog.

Where This Fits: Funding Arbitrage and Related Reading

A benchmark exists to be a reference point, not a trade signal by itself — but the gap it measures is exactly the raw material a funding arbitrage trade is built on, so it's worth being explicit about where the two connect and where they don't.

Cross-venue funding arbitrage calculator

A wide, correctly annualized spread between two venues in this benchmark is not the same thing as a profitable trade — opening and closing both legs costs round-trip takerA trader whose order executes against an existing order from another participant. This removes liquidity from the order book; the exchange applies its taker fee to that execution. fees, and moving collateral between a CEX and a self-custody DEX wallet costs a withdrawal fee that a raw spread comparison ignores entirely. Yieldo's funding arbitrage calculator nets those costs out of any spread you've spotted here, and the general funding rate guide and CEX vs DEX funding arbitrage guide walk through the mechanics in full if the concept itself is new to you. For a live, unfiltered board of current spreads across every venue and coin Yieldo tracks — not limited to the benchmark's popular-coin list — see the funding arbitrage index or the general funding rate hub.

Related: Hyperliquid vs Aster vs Lighter funding compared

The dataset summarizes observed API funding quotes after interval normalization, not settled payments or realized strategy returns. Hyperliquid, Aster and edgeX appear as separate venue groups when eligible data exists. The CEX group uses the median of available exchange values; membership follows active providers and data validity, not a permanent count of eight venues. Missing or rejected observations are not zero funding.

Risk warning: this benchmark is a measurement of historical funding rates, not a trading recommendation. Funding rates are volatile, can flip sign at the next settlement, and a wide historical spread is no guarantee that spread persists or that it clears trading and withdrawal costs. Perpetual futures trading involves substantial risk, including leverage-amplified losses and liquidation. Nothing in this article is financial advice. Do your own research and never risk more than you can afford to lose.

FAQ

What is the Funding Rate Benchmark and what does it actually measure?

The dataset summarizes observed API funding quotes after interval normalization, not settled payments or realized strategy returns. Hyperliquid, Aster and edgeX appear as separate venue groups when eligible data exists. The CEX group uses the median of available exchange values; membership follows active providers and data validity, not a permanent count of eight venues. Missing or rejected observations are not zero funding.

How do you make Hyperliquid's hourly funding rate comparable to a CEX's 8-hour funding rate?

Annualized rate (%) = raw decimal rate × (24 / interval in hours) × 365 × 100. Use the actual interval of each instrument. If the input is already a percentage, do not multiply by 100 again. This is a projection at an unchanged rate.

Why does the CEX group use the median instead of the average across exchanges?

Because a mean can be pulled off-center by a single exchange with a thin order bookA list of buy and sell orders with their prices and sizes. A trade occurs when compatible orders from two sides are matched. or a temporarily anomalous reading, even when the rest of the group agrees closely. A median resists that - one outlier exchange has to drag several others with it before it can move the reported figure. That is why the public field is named cex_median, not cex_average.

How far back does the data go, and why doesn't it go further?

The benchmark's history starts on 2026-07-05 - a hard floor set by when the underlying database table starts recording, not an editorial choice. There is no backtest, estimate, or reconstruction of funding rates before that date, and the feed publishes this exact date in its coverage_from field.

Is Aster included in the benchmark, and since when?

Yes, but on a separate and more recent timeline: Aster's funding history in this dataset starts on 2026-08-04, roughly a month after the general coverage_from date. Any average shown for Aster early on reflects however many days have accumulated since that later start, not a full window, and its chart line may not appear until it has at least two days of recorded history.

How can I download the raw data or cite this dataset?

Pull the JSON feed at yieldo.me/data/funding-benchmark.json or the CSV at yieldo.me/data/funding-benchmark.csv - both free, updated on a five-minute cache, licensed CC BY 4.0 with an attribution block built into the payload. The feed's own cite_as field reads: Source: Yieldo (yieldo.me) - real-time exchange data aggregated from official APIs.

Does a high or low reading in this benchmark mean I should open a trade?

Equal long and short exposure reduces directional price exposure, but does not eliminate basis risk, independent liquidation, execution failures or counterparty risk. Annualizing a current funding quote is a hypothetical projection, not realized profit or guaranteed net yield.

This article contains affiliate links. Yieldo may earn a commission at no extra cost to you — it never changes the fees or rates you see.

Written by Eugen Voyager — crypto analyst and founder of Telochain blockchain.

Last updated 02 October 2026.