Funding Alone Can Eat a 20x Position's Entire Liquidation Buffer — Before Price Moves

Two days ago I published the math on how long a leveraged position survives assuming a coin-flip and nothing but volatility hitting the liquidation barrier. That model has a gap: it treats margin as untouched until the barrier is hit. In reality, margin bleeds continuously from funding, paid every 8 hours regardless of whether your thesis is right, wrong, or hasn't resolved yet. So I pulled a live snapshot of funding rates across 861 USDT perpetual pairs off our own funding page and ran the numbers: how much of a position's liquidation buffer does funding eat during the exact window the position is statistically expected to survive?

What funding actually looks like across 861 live pairs

Snapshot taken 2026-09-25. Of 861 pairs, 552 had a positive funding rate (longs paying shorts), 42 were negative, and 267 sat at exactly zero. The distribution of how large the rate gets, in absolute terms, is heavily tailed:

The median pair is genuinely mild -- 5.5% annualized is a rounding error next to leverage risk. But "median" is doing a lot of work in that sentence. A quarter of live pairs already sit above 11% annualized, and the tail gets ugly fast: the 90th percentile is already near 30% annualized, and outliers like ONEUSDT show funding alone can run north of 500% a year on a single pair, in either direction depending which side you're on.

Turning that into margin cost, by leverage

Funding is charged on notional, not on margin -- but margin is what determines liquidation, and margin shrinks as leverage rises. So the same funding rate eats a bigger bite of margin per day the higher your leverage runs: at 20x, a 0.03%/day funding rate costs 20 times that as a share of margin, or 0.6%/day. I took each leverage tier's median survival time from the earlier model and asked: how much margin does funding alone consume over exactly that window, at the median, 90th, and 99th percentile rates?

Read the 20x row. The buffer -- the entire price move needed to get liquidated -- is 4.60% of margin. At the median funding rate, funding eats 1.56% of that over the 5.2 days the position is expected to live anyway: real, but not the main event. At the 90th percentile rate, funding eats 8.43% -- almost double the entire buffer -- with price not required to move a single tick. And it's not a rare pair: one in ten pairs in a live 861-pair snapshot sat at or above that rate. At 10x the gap is even starker: a 90th-percentile pair's funding bill (18.24%) is nearly double the 9.60% buffer across a 22.5-day median hold.

Even 100x isn't immune

The buffer at 100x is razor thin -- 0.60% -- but so is the median survival window, just 2.2 hours. I expected funding to be irrelevant at that horizon; it mostly is, at the median (0.14%). But at the 90th percentile rate, funding alone reaches 0.74% of margin in that same 2.2 hours -- more than the entire buffer. It takes an unusually hot pair to get there in two hours, but the dataset says roughly 1 in 10 pairs clears that bar right now.

What I'm actually checking now, on top of the old checklist

My working rule used to be leverage and volatility, full stop -- check the time to liquidation calculator, size for the buffer, done. This data adds a second question I wasn't weighting enough: which side of funding am I on, and how far from median is this specific pair? A position on a calm, near-zero-funding pair and the exact same position on a 90th-percentile pair aren't the same trade at the same leverage -- one has a buffer that's purely price risk, the other has an invisible second clock ticking against margin from the first funding settlement onward. Before sizing anything above 10x now, I check the live rate on that specific pair, not just the leverage math.