The answer up front: funding is positive most of the time because more people want leveraged long exposure than leveraged short exposure, and the excess demand has to be priced. That is a price, not a law.
The distinction matters. A law is something you can count on. A price moves with supply and demand — including down to zero, including below it. Below, this year’s settlements, period by period.
Start with the ratio: how much is “most of the time”?
The following comes from Binance’s public futures API, covering 1 January 2026 through the 2026-09-18 00:00 UTC settlement. Four symbols, 781 periods each, 260 days (read at 2026-09-18 07:13 UTC).
| Symbol | Positive | Negative | Exactly zero | Share positive | Year to date |
|---|---|---|---|---|---|
| BTCUSDT | 572 | 209 | 0 | 73.2% | 2.030026% |
| ETHUSDT | 534 | 247 | 0 | 68.4% | 1.208608% |
| BNBUSDT | 351 | 7 | 423 | 44.9% | 2.001652% |
| SOLUSDT | 391 | 390 | 0 | 50.1% | −1.030962% |
That table turns “funding is always positive” into four rather different conclusions. BTC is a little over seven in ten. SOL is a coin flip. And BNB produces no cash flow at all in more than half its periods.
Three sources of the positive bias
One: going long is easier, and more crowded
In spot markets you can buy and hold, but shorting means borrowing the coin first. Perpetuals removed that asymmetry — both directions need only margin. Habits move slower than mechanics, though: most retail flow arrives wanting upside, not downside.
The excess long demand pushes the contract above spot; once a premium exists, funding turns positive and longs start paying shorts. So funding is not something the market hands to shorts. It is rent that longs pay for leverage. For where the money physically comes from, see what a funding rate is.
Two: the formula carries an upward bias
Binance funding is not pure premium — it includes an interest-rate component, and you can read it straight off the public API: BTCUSDT reports 0.00010000, BNBUSDT reports 0.00000000.
That single field explains the strangest row in the table above: BNBUSDT has 423 periods that land exactly on zero. With a zero interest component, any premium inside the clamp band computes to precisely zero, while BTC’s positive interest component keeps it on the positive side even in calm conditions.
This is why the “baseline” can differ so much between symbols on the same venue and the same contract type. Extrapolating from BTC to BNB will be badly wrong. For the formula itself and the steps people miscalculate, see how the funding rate is calculated.
Three: bull markets multiply that bias
Year by year it is obvious. On our track record page, BTCUSDT accumulated 30.635% in 2021 and 11.980% in 2024 — but only 4.165% in 2022, and ETHUSDT managed just 0.787% that year.
Same mechanism, same symbol, an order of magnitude between years. The reason is not mysterious: when prices are running, more people want leveraged upside, and the rent gets expensive.
So why does it turn negative?
Negative funding is the same mechanism in reverse: shorts crowd out longs, the contract slips below spot, and shorts start paying longs.
In this year’s data that is hardly rare:
- SOLUSDT: 390 negative periods out of 781, cumulative −1.030962% year to date. Put another way, anyone holding SOL perpetual shorts has been a net payer on funding this year.
- ETHUSDT: 8 negative periods out of 52 in September so far, low of −0.01081%.
- BTCUSDT: just 1 negative period over the same 52, low of −0.00015% — effectively noise.
There is a usable conclusion in that contrast: the more speculative the symbol, the more often funding crosses between positive and negative. BTC’s funding behaves like a low flat baseline; SOL’s behaves like a wave repeatedly cutting through zero. Whether you can run the trade in reverse when funding is negative is worked out in what negative funding means.
The part that matters most: positive is not the same as enough
This is the section most articles leave out.
BTCUSDT was positive in 73.2% of periods this year, which sounds dependable. But add all 781 together and you get 2.030026%. Spread across 260 days, that is 2.85% annualised.
In the same week, the Federal Reserve lifted the federal funds target range to 3.75%–4.00%.
So: this year to date, the gross return on the funding leg has trailed the short-term risk-free rate — and that is before trading costs.
Frequency and magnitude are different things. Funding can rarely turn negative and still be too thin to be worth running. That is exactly this year’s shape: not many negative periods, but an average of roughly 0.0026% per period.
What to watch
No forecasts — three things you can count yourself:
- Share of positive periods. BTC is at 73.2% this year. If that drops under 60%, negative periods are getting denser, not merely shallower.
- Average per period. BTC’s 30-day average is 0.006816%. It is more sensitive than the share, and it moves first when magnitude collapses.
- ETH and SOL move before BTC. In September, BTC had 1 negative period, ETH had 8, SOL had 22. If you want early warning of a turn, watch those two.
All of these sit on our funding rate data and track record pages, ETHUSDT’s 0.787% in 2022 included — funding thinning toward zero or going negative in flat and bear markets is something this approach genuinely runs into, and we have not hidden it. For how the whole process runs, see how it works.
Sources: Binance public futures API (per-period funding history, premiumIndex interest component; read at 2026-09-18 07:13 UTC, latest settlement 2026-09-18 00:00 UTC); federal funds target range per the Federal Reserve FOMC statement of 16 September 2026; yearly totals on our track record page, on-site snapshot timestamp 2026-08-25 13:42 UTC.