Direct answer
In September 2026 BTC, ETH and SOL each settled 90 times on Binance and on OKX, with complete records. A 10,000 USDT long held all month paid 45.62 (Binance) and 44.79 USDT (OKX) on BTC, 40.60 and 40.94 USDT on ETH, and only 26.25 and 18.05 USDT on SOL. No interval on any of the three exceeded +0.0100%; estimating with +0.01% × 90 intervals would have nearly doubled the funding on a BTC long.
Completeness: 90 settlements for every contract
The period runs from 1 September 00:00 to 1 October 00:00 2026 in Taiwan time (UTC+8). All three contracts settled every 8 hours on both exchanges throughout, so 90 settlements were expected and 90 records were retrieved for each, with no gap longer than 8 hours — so all three can be compared across exchanges.
Under the report method completeness comes before any statistic: if one exchange were missing a few days, its average and total would describe an incomplete sample and could not be ranked against the other.
| Contract | Exchange | Settlements expected | Records | Longest gap |
|---|---|---|---|---|
| BTC | Binance | 90 | 90 | 8 hours |
| BTC | OKX | 90 | 90 | 8 hours |
| ETH | Binance | 90 | 90 | 8 hours |
| ETH | OKX | 90 | 90 | 8 hours |
| SOL | Binance | 90 | 90 | 8 hours |
| SOL | OKX | 90 | 90 | 8 hours |
A month of funding: what a 10,000 USDT long paid
Adding up the actual rate at every settlement gives what the same 10,000 USDT long paid from 1 September to the end of the month; a short received the same amount. BTC and ETH differed by less than 1 USDT between the exchanges, SOL by 8.21 USDT.
SOL's gap has two parts: OKX had 34 negative intervals against Binance's 29, and OKX's average rate per interval was about 0.0009 percentage points lower. For a SOL long, September's funding was lower on OKX; for a short, the reverse.
| Contract | Binance, 10,000 USDT long | OKX, 10,000 USDT long | Difference |
|---|---|---|---|
| BTC | pays 45.62 USDT | pays 44.79 USDT | Binance 0.82 USDT more |
| ETH | pays 40.60 USDT | pays 40.94 USDT | OKX 0.35 USDT more |
| SOL | pays 26.25 USDT | pays 18.05 USDT | Binance 8.21 USDT more |
BTC: 85 positive intervals, capped at +0.0100%
BTC was positive in 85 intervals on Binance, with a longest run of 60 — twenty full days of longs paying; on OKX it was positive 80 times and negative 10. Both averaged about +0.0050% per interval, roughly half the +0.0100% baseline.
The lowest interval was −0.0026% on Binance (28 Sep 16:00) and −0.0027% on OKX (6 Sep 00:00), both shallow: at most about 0.27 USDT received on a 10,000 USDT long in one interval.
| Measure | Binance | OKX |
|---|---|---|
| Average per interval | +0.0051% | +0.0050% |
| Median | +0.0048% | +0.0050% |
| 10th / 90th percentile | +0.0010% / +0.0094% | −0.0002% / +0.0097% |
| Highest (intervals at it) | +0.0100% (8) | +0.0100% (9) |
| Lowest (UTC+8) | −0.0026% (28 Sep 16:00) | −0.0027% (6 Sep 00:00) |
| Positive / negative intervals | 85 / 5 | 80 / 10 |
| Longest run of positive rates | 60 intervals | 52 intervals |
ETH: the deepest negative rate came on 11 September
ETH averaged +0.0045% per interval on both exchanges, with medians of +0.0052% (Binance) and +0.0045% (OKX). Positive intervals were almost the same: 81 on Binance, 82 on OKX.
The month's most negative interval came at the same moment on both: 11 Sep 00:00 UTC+8, −0.0108% on Binance and −0.0060% on OKX. Same moment, same direction, nearly twice the depth — one reason one exchange's rate cannot stand in for the other's.
| Measure | Binance | OKX |
|---|---|---|
| Average per interval | +0.0045% | +0.0045% |
| Median | +0.0052% | +0.0045% |
| 10th / 90th percentile | 0.0000% / +0.0090% | 0.0000% / +0.0084% |
| Highest (intervals at it) | +0.0100% (6) | +0.0100% (6) |
| Lowest (UTC+8) | −0.0108% (11 Sep 00:00) | −0.0060% (11 Sep 00:00) |
| Positive / negative intervals | 81 / 9 | 82 / 8 |
| Longest run of positive rates | 35 intervals | 26 intervals |
SOL: about a third of settlements were negative
SOL had negative rates most often of the three: 29 intervals on Binance, 34 on OKX. Both 10th percentiles were about −0.0040%, meaning a tenth of settlements were more negative still, while the 90th percentile reached +0.0100%.
The deepest interval was 2 Sep 00:00 on both, −0.0088% on Binance and −0.0115% on OKX, and OKX also had a run of 9 negative intervals — three full days. On average OKX settled at +0.0020% per interval and Binance at +0.0029%.
| Measure | Binance | OKX |
|---|---|---|
| Average per interval | +0.0029% | +0.0020% |
| Median | +0.0028% | +0.0016% |
| 10th / 90th percentile | −0.0040% / +0.0100% | −0.0041% / +0.0100% |
| Highest (intervals at it) | +0.0100% (18) | +0.0100% (16) |
| Lowest (UTC+8) | −0.0088% (2 Sep 00:00) | −0.0115% (2 Sep 00:00) |
| Positive / negative intervals | 61 / 29 | 56 / 34 |
| Longest run of positive rates | 21 intervals | 20 intervals |
Why the highest rate stopped at +0.0100%
Across three contracts and two exchanges, no September interval exceeded +0.0100%, and that exact figure came up repeatedly. It is not a coincidence: under Binance's published formula, whenever the premium index is between −0.04% and 0.06% the rate equals the 0.01%-per-8-hours interest component, and OKX's public data also show an interest component of 0.01%.
In other words, throughout September contract prices never traded clearly above spot, and crowding among longs never pushed the rate past the baseline; the movement happened below 0.01%, and sometimes below zero.
How far off a +0.01% estimate is
A common estimate of holding cost assumes +0.01% per interval at three intervals a day: 90 intervals, or 90 USDT a month on a 10,000 USDT long. Against September's actual settlements — 45.62 USDT on BTC, 40.60 USDT on ETH and 26.25 USDT on SOL at Binance — that estimate is far too high.
+0.01% is the top of the range in calm markets, not the average. To estimate holding cost, the average of recent actual settlements comes closer to what you would really pay than the baseline rate does.
What this report cannot tell you
It cannot forecast October: funding is set afresh each interval, and September's average describes September only. Nor does it include fees, slippage or price gains and losses, or say which exchange suits you.
The conversion uses a fixed 10,000 USDT position; in practice each interval's funding is charged on the position's value at settlement and moves with the price. For your own position, enter the rates in the funding cost calculator.
Official sources
The data are Binance's and OKX's published funding histories — the rate actually applied at every settlement — retrieved on 2026-10-07, and the statistics follow the site's monthly funding report method. Both histories are public, so anyone can recompute every figure on this page for the same period. Recorded 2026-10-07.
- Binance developer docs: USDⓈ-M futures market data (funding history, funding limits, contract specs)
- OKX API documentation: instruments, funding rate and funding rate history
- Binance: Introduction to Binance Futures Funding Rates
- OKX: Perpetual Futures Funding Fee FAQ
Keep reading
Frequently asked questions
Where do these figures come from?
From Binance's and OKX's published funding histories — the rate actually applied at each settlement — not from the site's own predictions. Both histories are public, so the same period can be recomputed by anyone.
Why is the month cut in Taiwan time?
The site writes its reports in Taiwan time: September runs from 1 September 00:00 to 1 October 00:00 UTC+8, exactly 90 settlements. Cutting the month in UTC would move one settlement at each end, so the totals would differ slightly.
Can September's results predict October?
No. A report records settlements that have happened, and funding is set afresh every interval by the market at the time, so next month can look entirely different.
Why convert everything to a 10,000 USDT long?
A fixed position lets different contracts and exchanges be compared directly. Scale by notional for your own size; a short is the same amount in the other direction.