Funding rate history data

878,142

Settlement records

Every period is auditable

450

Pairs covered

Binance perpetuals

2020–2026

Span

As of 2026-09-14

86.0%

BTC periods that were positive

7,344 periods

Read this before you read any number below

Every percentage on this page is a gross market rate. Not one of them is net of trading fees on either leg, cost of capital, slippage, or the cost of exiting. They describe what the market paid, not what anyone received.

That distinction is the single most common way funding-arbitrage numbers mislead. A headline built on gross spreads and a number built after four cost lines are deducted can differ by a wide margin, andneither one is technically a lie. So the useful question is never "what is the APR" — it is "which costs were taken out before you printed it". Ours are itemised onhow it works, and the realised figures are onperformance.

The question this data answers

The whole case for funding rate arbitrage rests on one thing:whether perpetual funding is positive more often than not, over a long enough sample. Positive funding means longs pay shorts, and the short leg of a delta-neutral position is the side being paid. That is not a judgement call — it is a count, and it needs a sample long enough to include a bear market.

The two rows below are that sample. They are the only pairs in this dataset with more than two years of continuous history; the other several hundred only start in August 2025, and using them to describe a multi-year trend would be inventing a trend.

The last column needs the caveat spelled out. "Naive APR" is the average rate per period multiplied by three periods a day and then by 365. It deductsno fees, no cost of capital, no slippage, assumes perfect entry timing, and does not compound. Its only job is to turn "0.01% per period" into a magnitude a human can feel. Anything actually achievable is lower — by how much, and why, is set out onrisk and boundaries.

Month by month: how that line was actually walked

The table above compresses six and a half years into one row, and an average hides the path. The chart below spreads the same data out by month. Switch to "Per month" and the negative months become visible — in those, the short side of the trade was not being paid, it was paying. That is the part "funding is positive over the long run" quietly leaves out.

Year by year: how many periods were positive

The jump in the "pairs" column is not a data error. Before August 2025 only BTC and ETH were collected continuously; after that the coverage widened to several hundred. So the 2020–2024 rows describetwo majors and the 2025–2026 rows describe the whole perpetual market. The two segments are not directly comparable — which is exactly the point of the next section.

A fact that is easy to bury: majors and the market are not the same thing

Look at "average per period" in the last two rows above — they are negative. Averaged across several hundred perpetual contracts, both 2025 and 2026 sit below zero. Over the same stretch, BTC and ETH remain positive on a long-run average, with roughly 86% of periods positive.

The two numbers do not contradict each other, and together they say something specific:"funding is positive over the long run" holds reliably only on the most liquid pairs. Extend it across the long tail and it stops being true. A large number of newly listed small-cap perpetuals sit in persistently negative funding, where the short side does not collect — it pays.

This is why using a market-wide average as the benchmark for a funding strategy is wrong, and whythis strategy only takes positions on major pairs. A table that reports the market-wide average without splitting it apart hides precisely this.

Last 30 days: is the market paying or charging right now

The window runs 30 days back from the last settlement in the database (2026-09-14), counting only pairs with at least 60 settlements inside it. 450 pairs qualified, of which376 were cumulatively positive (83.6%). A further 4 pairs count toward those totals but are left out of the lists below — their tickers are non-ASCII meme coins, irrelevant to the point this page is making.

The right-hand table is the part of this page most worth reading. The double-digit positives on the left are appealing, but the same window contains pairs down near −80%. Pick the wrong direction, or a pair that funds negative structurally, and the loss is just as real.Nothing in "funding rate arbitrage" guarantees you are on the receiving side — you are paid only if you picked correctly.

Methodology

Next