Context and references.

Binance BTCUSDT perpetual funding averaged about -0.0013% per 8-hour settlement over 1-11 May 2026 and about +0.0029% over 12-19 May (24 settlements ranging from -0.0042% to +0.0072%, five of them negative). Binance's funding-history page lists every settlement.

For context, over the twelve months to mid-May 2026 the weekly average had a median of about +0.0042%, a 75th percentile of +0.0059% and a 90th percentile of +0.0067%, and no single BTCUSDT settlement went above +0.01%. February and March 2026 averaged slightly negative. May 2026 sits at the low end of an already cool range — no sign of leverage building. (Source: Binance funding-rate history, own calculation.)

Cross-venue comparison is best done on an aggregated view such as Coinglass, which puts Binance, OKX, Bybit and others side by side. Small gaps between venues usually reflect per-venue order book conditions rather than market disagreement.

For position sizing, a reading this close to zero means carry cost is not the issue; direction is. Sustained funding well above +0.03% — as in early March 2024, when BTCUSDT peaked at +0.0881% on 5 March — is the environment where crowded longs become the risk.

What the funding figure actually measures.

That figure is a rate per settlement interval — not an annual rate, not a daily one. Binance settles BTC perpetual funding every eight hours, so a reading like +0.0039% is what a long pays a short three times a day on notional, and only if the position is open at the settlement timestamp. Open a long just after one settlement and close before the next and you pay nothing at all. That is why a number which quietly bleeds a swing position is irrelevant to someone flat by the end of the session.

Cross-venue comparison carries a trap worth knowing. A gap between Binance, OKX, Bybit and dYdX only means something if all four are quoted on the same interval — and not every venue on that list settles on an eight-hour cycle. Coinglass normalizes this on its comparison view, but individual venue pages report their own native interval. Read the interval label before concluding one venue is cheaper than another, because getting this wrong makes a venue look several times more or less expensive than it is.

The last thing worth recording: a weekly average hides its own distribution. A week of flat settlements near +0.003% and a week with two +0.03% prints and roughly zero everywhere else both average out to about the same place, and they are entirely different risk environments. That is precisely why the percentiles above are quoted next to the mean. If your weekly notes only capture the average, you have thrown away the part that actually precedes deleveraging.

Crypto assets are volatile and not suitable for every investor. This page is editorial analysis, not financial advice.