While re-pulling this year’s settlement data we also counted the interestRate field across Binance’s USDT-margined premiumIndex endpoint.
Read at 2026-09-20 14:11 UTC, 908 symbols:
| Interest rate | Symbols | Share |
|---|---|---|
| 0.01% / 8h | 669 | 73.68% |
| 0 | 237 | 26.10% |
| 0.005% / 8h | 2 | — |
Four of the 237 are quarterly delivery contracts, which never charge funding at all, so 233 of them are genuine perpetuals. The two on 0.005% are OPENAIUSDT and ANTHROPICUSDT.
This is not trivia. That field directly determines where a contract’s funding reading gets pinned.
It is the centre of the clamp
Binance’s published formula:
funding rate = premium index + clamp(interest rate − premium index, −0.05%, +0.05%)As long as |interest rate − premium index| ≤ 0.05%, the clamp does not bind, the two terms cancel, and funding comes out exactly equal to the interest rate. That is the dead band.
The two classes of contract have their dead bands at different heights:
| Interest rate 0.01% | Interest rate 0 | |
|---|---|---|
| Premium range inside the band | −0.04% to +0.06% | −0.05% to +0.05% |
| Funding reading inside the band | exactly 0.01% | exactly 0 |
| To exceed 0.01% | premium > +0.06% | premium > +0.06% |
In other words: a contract with a 0.01% interest rate comes with a floor — as long as the premium is not too negative, it still collects 0.01%. A contract with a zero interest rate has no such floor.
BNBUSDT is the cleanest example
197 of the 233 are TradFi underlyings (US and Hong Kong and Korean equities, gold and silver, crude, leveraged ETFs and so on), a class whose pricing logic differs from crypto anyway.
But among the thirty-odd that remain are BNBUSDT and BNBUSDC, along with ETHBTC, WAVESUSDT and a batch of mid- and small-cap coins. BNB is one of Binance’s own major symbols, not some fringe listing.
Classifying this year’s 788 periods (2026-01-01 through 09-20 08:00 UTC) by reading:
| Symbol | Interest rate | Periods exactly 0 | Periods exactly 0.01% | Single-period max |
|---|---|---|---|---|
| BTCUSDT | 0.01% | 0 | 54 | 0.010000% |
| ETHUSDT | 0.01% | 0 | 42 | 0.010000% |
| SOLUSDT | 0.01% | 0 | 102 | 0.010000% |
| BNBUSDT | 0 | 423 (53.68%) | 0 | 0.032423% |
That table says two things at once:
- More than half of BNB’s settlement periods this year came out at exactly zero, and not one landed on 0.01%. That is not coincidence — its dead band sits at zero.
- BTC, ETH and SOL have no zero periods at all, yet each has dozens to over a hundred sitting exactly on 0.01%. Also not coincidence.
The instant of the read makes it more vivid: BNBUSDT’s mark price was 0.0489% above its index — the perpetual was at a premium — and its predicted funding was 0.000000%. At the same moment BTCUSDT was 0.0458% below its index, and its predicted funding was 0.008140%.
The two symbols’ premiums had opposite signs, and the higher funding belonged to the one at a discount.
So has BNB collected less this year
No. This is the most interesting part of the data:
| Symbol | Year to date | Annualised |
|---|---|---|
| BTCUSDT | +2.091425% | 2.910% |
| BNBUSDT | +2.059435% | 2.865% |
| ETHUSDT | +1.265904% | 1.761% |
| SOLUSDT | −0.960962% | −1.337% |
BNB collected nothing on half its periods and still finished the year almost level with BTC.
The reason is in the last column of the previous table: BTC’s highest single period this year was 0.010000%, never once above baseline, while BNB’s highest reached 0.032423% — more than three times that. A contract with a zero interest rate has no floor, but it also has no layer flattening the premium into 0.01%. When longs genuinely crowd, the reading goes straight up.
One more difference: BNB had only 7 negative periods out of 788 this year, against 209 for BTC and 390 for SOL. A zero interest rate pulls the mildly negative stretch up to zero as well.
What it means in practice for arbitrage
Three things.
One: check this field before comparing funding across symbols. Putting two contracts’ current readings side by side is systematically unfair to the zero-baseline class — inside the dead band it reads 0 while its counterpart reads 0.01%. It is the same category of error as checking the settlement interval before comparing funding across exchanges: the readings are correct, the conventions are not the same.
Two: the cash flow is lumpier. Half the periods pay nothing and the other half pay more. The total is similar, but for anyone reconciling period by period the variance is much higher.
Three: the exchange can change this parameter. Like funding caps and settlement intervals, it is an exchange-adjustable contract parameter — the same endpoint shows 789 symbols carrying custom caps and intervals. Rule parameters being changed is one form of counterparty risk.
What it means for a balanced two-leg position
It does not mean switch symbols. BNB’s year-to-date total is roughly level with BTC’s, while its depth, margin brackets and delisting risk are all different, and symbol selection in this strategy is a source of risk rather than a source of return.
What it does mean is this: every funding reading you look at carries a set of parameters you are not looking at. Interest rate, settlement interval, caps — if any one of the three differs, two readings are not in the same coordinate system. All three are in the public API. Nobody reads them.
What to watch next
- BNBUSDT’s share of zero periods. It is more sensitive than the cumulative figure: a rising share means the premium is converging toward zero.
- Whether the list of 233 changes. It is an exchange-adjustable parameter, and a change does not necessarily get its own announcement.
- The 0.005% on OPENAIUSDT and ANTHROPICUSDT. They are the only two in the whole market on that setting, so it is worth watching whether it becomes a third class.
To finish the thought properly: funding goes to zero and turns negative — the 0.787256% ETHUSDT line for 2022 on the track record page is how that happened, and SOLUSDT is at −0.960962% year to date. BNB had only 7 negative periods this year; that is this year’s reading, not a promise about any other.
Sources: the interestRate distribution, mark and index prices and predicted funding come from Binance’s public USDT-margined futures premiumIndex endpoint, the TradFi-versus-crypto classification from underlyingSubType in exchangeInfo, and the funding caps and settlement intervals from fundingInfo, all in the same read (2026-09-20 14:11 UTC); this year’s 788 settlement periods come from fundingRate, covering 2026-01-01 through 09-20 08:00 UTC; the funding formula and the interest-rate convention are Binance’s published definitions; yearly totals on the track record page, site snapshot as of 2026-08-25 13:42 UTC. Exchanges can change any of these contract parameters at any time. Past performance does not indicate future results.