A trader I know held a 10x long on ETH for about six weeks this spring. His entry and his exit were within four dollars of each other. He still came out of that position down roughly 12% on his account.
Nothing went wrong with the trade. He just never opened the funding tab.
Perpetual futures don't expire, and that's the whole appeal. No settlement date, no quarterly rollover, no closing a position you still believe in because a contract ran out of calendar. But something has to keep a contract with no expiry glued to the spot price, and that something is funding.
Every eight hours, money moves between the two sides of the market. When more size wants to be long than short, longs pay shorts. When the crowd is short, it flips. Three payments a day, every day, whether or not the price does anything at all.
Most traders know this. Almost none of them price it.
The baseline rate on major venues is 0.01% per interval. That's about 11% a year on the position — irritating, survivable. At 0.05%, a level that shows up regularly whenever a market gets hot, you're paying somewhere near 55% a year.
I've watched crowded alt perps sit near 0.3% per interval for days at a stretch, which annualizes past 300%. At that point you're not trading the asset. You're renting it at rates a credit card company would be embarrassed to quote.
Here's the part that does the real damage: funding is charged on notional, not on your margin.
Post $10,000, open a $100,000 position at 10x. At a mild 0.01% you pay $10 every eight hours. Thirty dollars a day. Around $900 a month.
That's 0.9% of the position and 9% of your actual money. Annualize it and your account is paying more than 100% a year to hold a trade you've mentally filed as free.
Leverage doesn't just multiply your exposure. It multiplies the rent.
The bill arrives at the worst possible time
Funding isn't random. It's a crowding gauge — the rate goes positive and stays positive precisely when everyone wants the same side of the trade.
Which means the cost of holding your long peaks in the exact week your conviction peaks. The room is most expensive when it's most full. You feel smart, the tape agrees with you, and you're quietly paying three times a day for the privilege of agreeing with everyone else.
Then there's where the money comes from. Funding is deducted from your margin, not billed separately, so every payment nudges your liquidation price a little closer to the current price. A position that was comfortably collateralized on Monday can be sitting on a thinner buffer by Friday without the chart moving an inch.
That's the mechanism behind a specific kind of bad day: the trader who was right about direction, wrong about nothing in particular, and got taken out by a wick that would have missed him at entry.
Somebody is collecting this
The most useful thing I can tell you about funding is that it isn't a fee the exchange invented. It's a transfer. Someone is on the other end.
That someone runs the basis trade — long spot, short the perp, no directional exposure, collecting funding from leveraged longs. In 2025 that trade returned somewhere around 19% annualized with drawdowns under 2%. It's a boring, market-neutral yield strategy, and it's funded almost entirely by retail traders paying to stay leveraged through flat markets.
That yield isn't clever. It's your funding tab, redistributed.
What to do about it
Put funding in the strategy instead of the footnotes. Before you enter, work out what the position costs per day at the current rate, and ask whether the move you're expecting covers it inside the holding period you're expecting. Three weeks at 0.03% needs about 1.9% just to get back to flat — before fees, before slippage.
Then set a funding threshold that reduces or closes the position regardless of your view. I'd rather exit a correct trade that got too expensive than watch a right call bleed out at three payments a day.
And if you don't want to model any of it, one rule covers most of the damage: don't hold high-leverage perps through a market that isn't moving. Time is not neutral in this product. It's a cost with a schedule.
None of this requires judgment. It's arithmetic, on a timer, which is exactly the sort of thing a person watching a chart handles badly and a system handles without being asked. Our strategies price funding and fees into the backtest rather than discovering them on the statement — full data and methodology are at v33systematic.com.