Overview
Funding rates are periodic payments exchanged between the long and short holders of a perpetual futures contract to keep its price aligned with the underlying spot or index price. Unlike traditional futures, perpetual contracts never expire, so there is no settlement date that forces convergence. The funding mechanism substitutes for that: it answers what it costs to hold a leveraged perpetual position, and which side of the market is paying to stay in it. Traders, risk teams, and researchers use funding rates to measure the cost of carry, gauge crowding and sentiment, and compare the economics of perpetuals across venues. Coin Metrics collects funding rates from perpetual futures markets across its exchange coverage universe and serves them as a per-market time series over the HTTP endpoint/timeseries/market-funding-rates.
Realized, not predicted. This page covers the realized funding rate: the rate actually calculated over the previous funding interval and used to determine the funding payment. Some exchanges also publish a forward estimate of the next rate, which they may call the predicted, real-time, or next funding rate. For that series, see Market Funding Rates Predicted.
At a Glance
| Data type | Entities | Frequency / cadence | Unit | Primary endpoint | Coverage |
|---|---|---|---|---|---|
| Realized perpetual-futures funding rates | Markets (perpetual futures) | Event-driven. One observation per market at the end of each funding interval (commonly every 8 hours, but venue-specific, for example hourly on some venues) | Funding rate as a decimal fraction over the period (for example 0.0010 = 0.10%). The period and interval are returned as separate duration fields | /timeseries/market-funding-rates | 🔗 |
Schema
One observation is the realized funding rate for one market at the end of one funding interval. Every observation carries the fields below. The columns are the response schema for/timeseries/market-funding-rates.
Conventions. The
rate is returned as a JSON string to preserve precision. Timestamps are UTC ISO-8601 with nanosecond resolution: time is the exchange-reported event time (the end of the funding interval), and database_time is when Coin Metrics stored the observation. The period and interval fields are HH:MM:SS durations. The rate is a decimal fraction over the period, not a percentage or an annualized number (see Period, interval, and annualization).Methodology
Funding rates are reported by each exchange under its own mechanism design, and Coin Metrics records them as a harmonized per-market time series rather than re-deriving the payments. The mechanics below cover how the data is collected, how the rate is signed and denominated, how theperiod and interval differ, how observations are timestamped, and how duplicates are removed.
Collection
Coin Metrics collects realized funding rates from the perpetual futures markets of the exchanges in its coverage universe, using whichever method a venue supports:- Streaming. Where an exchange pushes funding over a websocket, Coin Metrics records each update as the exchange sends it.
- Polling. Where funding is available only over a REST endpoint, Coin Metrics polls it and records the current value for each market.
- Historical backfill. History is loaded from each exchange’s historical funding endpoint so that the series extends back to the start of coverage.
Sign convention and units
Therate is a decimal fraction that represents the return over one period, so a rate of 0.0010 is 0.10% over that period. The sign follows the direction of the payment:
- When the perpetual price trades above the underlying spot or index price, the rate is positive and long holders pay short holders.
- When the perpetual price trades below it, the rate is negative and short holders pay long holders.
Period, interval, and annualization
Two duration fields describe the timing of the rate, and they are not the same thing:periodis how long the rate applies to, that is, how often funding is calculated and exchanged. Many venues use an 8 hour period, some use 1 hour.intervalis the window of input data used to compute the rate. It is often equal to theperiod, but not always. Some venues measure inputs over a longer window than the payment period, and some calculate funding continuously, in which case theintervalis set to00:00:00.001by convention.
rate of 0.0001 annualizes to roughly 0.0001 * (8760h / 8h), or about 10.95% per year.
Timestamps and timeseries normalization
Exchanges differ in how they report the timestamp for a funding rate. Many publish funding as a time series of equally spaced points whose timestamps mark when the rate and payments are calculated. Others publish only a snapshot of the current rate at query time, which requires transformation to build a series. Regardless of how a venue reports it, Coin Metrics stores funding as a time series with one observation at the end of each funding interval, and returns that exchange-reported event time in thetime field. The separate database_time records when the observation was stored, so the gap between the two measures collection lag.
Deduplication
Coin Metrics runs redundant collectors for resilience, so the same observation can be produced more than once. Observations are keyed by market and event time, so two observations for the same market at the sametime collapse to a single stored point. This holds across the streaming, polling, and historical backfill paths, which lets them overlap without creating duplicate rows.
Accessing the Data
Funding rates are available over HTTP for historical queries and for the latest observations. There is no websocket stream for this data type. Query one or more markets over a time range at/timeseries/market-funding-rates, or use limit_per_market for the most recent observations. The markets parameter accepts a comma-separated list or wildcard patterns such as binance-*, *USDT-future, or *-future, so you can query many markets in one call.
- Python Client
- Shell
- Python
/timeseries/market-funding-rates.
Examples
The examples below show the latest observations for two markets with different funding conventions. Therate is returned as a JSON string.
Example: 8-hour funding
The latest funding rates forbinance-BTCUSDT-future, where the rate applies every 8 hours and is calculated over an 8 hour interval. Run this query.
Example: hourly funding measured over a longer window
The latest funding rates forkraken-PF_XBTUSD-future, where the rate applies every hour (period 01:00:00) but is measured over a 4 hour window (interval 04:00:00). This shows why period and interval are separate fields. Run this query.
Coverage
https://coverage.coinmetrics.io/market-funding-rates-v2Usage
- Current cost of carry. Use
limit_per_marketfor a quick “latest N” look at a market’s funding, or a wildcard such as*-futureto scan many markets at once. - History and backfill. Use a
start_time/end_timerange with.parallel(time_increment=…)to pull long histories efficiently. - Cross-venue comparison. Convert rates to a common basis before comparing, using the
periodfield to annualize (see Period, interval, and annualization). - Positioning and sentiment. Read the sign and magnitude of the rate to see which side is paying and how strong the imbalance is.
- Derivatives context. Combine funding with Market Open Interest and Market Liquidations on the same markets to study leverage build-up and unwind, and with the Funding Rate Metrics for a coverage-wide view.
Limitations
- Conventions vary by exchange. Each venue sets its own funding mechanism,
period, andinterval, so raw rates are not comparable across venues until they are put on a common basis. Annualize using theperiodbefore comparing (see Period, interval, and annualization). - Realized rate only. This series is the rate calculated over the previous interval. For the forward estimate of the next rate, use Market Funding Rates Predicted.
- Zero and near-zero rates are valid. Some venues can report a funding rate of
0. For example, Bitfinex does not require a funding payment when the average spread over the funding period stays within a small band, so long stretches of0are expected rather than missing data. - Snapshot venues are transformed into a series. For exchanges that publish only a current-rate snapshot rather than a time series, Coin Metrics reconstructs the series with one observation per funding interval (see Timestamps and timeseries normalization).
FAQ
What is the difference between the period and interval fields?
They describe two different things. The interval is the window of input data used to calculate the rate. Many exchanges use a 1 hour or 8 hour input window, and for exchanges that calculate funding continuously the interval is set to 00:00:00.001 by convention. The period is how long the resulting rate applies to, that is, how often funding payments are calculated and exchanged between long and short holders. The two are often equal (for example 8 hours and 8 hours), but not always: a venue can report an hourly rate measured over a 4 hour window, in which case period is 01:00:00 and interval is 04:00:00.
How can I annualize the funding rate to compare across exchanges?
Because exchanges use different periods, scale each rate to a common basis using theperiod field. Since funding payments are applied to the position value and are not compounded, annualizing is a simple multiplication by the number of periods in a year:
For example, an 8-hour rate of 0.00005 (near the Binance level in the example above) annualizes to 0.00005 × (8760h / 8h), or about 5.5% per year.
What determines the frequency of funding rates data?
A market produces a new observation at the end of each funding interval, so the frequency follows the venue’s funding schedule (commonly every 8 hours, hourly on some venues). See the coverage page for the markets available per exchange.Why are there so many funding rate values of 0 for Bitfinex?
Bitfinex allows a 0 funding rate. Under Bitfinex’s funding rate methodology, a funding payment is only required when the average spread over the funding period is greater than 0.05% or less than -0.05%. When the average spread stays within that band, no payment is required and the rate is 0.
Related
- Market Funding Rates Predicted: the forward estimate of the next funding rate for a market.
- Funding Rate Metrics: the open-interest-weighted aggregate and cumulative funding rate across markets, by asset and exchange-asset.
- Market Open Interest: outstanding contracts on the same derivatives markets.
- Market Liquidations: forced closes on the same derivatives markets.

