A 0.1% round-trip trading fee, a 0.01% funding rate every eight hours, a 30-day holding period, and a $100,000 position: those four numbers produce a surprising comparison. The fee is about $100. Funding is about $270 if the rate stays constant. The position can lose nearly three times as much to funding as to entering and exiting.
That’s easy to miss in a perpetual futures backtest. Fees appear on trades, right beside the fills. Funding arrives on a separate schedule, often as a small rate that looks harmless in a data table. But a small rate charged repeatedly against a large notional position is a real cash flow.
How do you calculate funding cost on a perpetual?
For a simple estimate, multiply the position’s notional value by each funding rate that applies while the position is open, then add the payments:
Funding cash flow ≈ position notional × funding rate
With $100,000 of notional and a positive 0.01% rate, a long paying funding owes $10 for that interval. If the rate repeats every eight hours for 30 days, there are 90 intervals: $10 × 90 = $900, or 0.9% of notional. The earlier $270 example uses a 0.003% rate per interval, which totals 0.27% over 90 intervals.
Funding direction matters. On many venues, when the rate is positive, longs pay shorts; when it’s negative, shorts pay longs. The precise convention, settlement interval, and notional basis depend on the contract and venue. A backtest should use the historical rate and the position actually held at each settlement, following that venue’s rules.
Why can the average funding rate give the wrong answer?
The position may not be open for every settlement, and its notional may change. A strategy that enters shortly before a settlement and exits afterward owes a payment even if it held the trade for only minutes. Another strategy might hold through most of the period but avoid the settlement entirely. Charging funding once per day, or prorating by hours held, can assign the wrong cash flow.
There’s a second trap: the average rate can hide when the strategy was exposed. Suppose funding is usually 0.003% per interval, but spikes to 0.06% during four settlements when your strategy tends to hold a crowded long. A broad monthly average may look modest, while the strategy’s position-weighted bill is large. Calculate against the positions and settlement timestamps in the simulation; don’t multiply the average rate by days held and call it done.
| Backtest shortcut | What it misses | Better treatment |
|---|---|---|
| One average rate for the month | Rate spikes during the strategy’s exposure | Apply each historical settlement rate to the open notional |
| Funding prorated by hours held | Discrete settlement timing | Charge only settlements the position was subject to |
| Rate applied to entry price | Notional can move as price and size change | Follow the venue’s funding basis and contract rules |
What should a funding-aware backtest record?
For each settlement, keep the timestamp, rate, position side, contract quantity, notional used by the venue, and resulting cash flow. That record makes it possible to inspect a costly payment instead of trusting a single total-cost number. It also catches the embarrassing sign error: a funding series stored from the venue’s perspective can be interpreted backward if the backtest assumes the opposite payment convention.
Compare the funding ledger with the strategy’s position timeline. If the position was flat at settlement, there should be no funding charge. If it held a $50,000 long through a 0.02% positive interval under a long-pays convention, the expected payment is $10. Small examples like this are useful accounting checks before you trust a multi-year curve.
And look at costs by regime. Funding can turn a strategy’s apparent edge into a bet that the same crowded side stays cheap to hold. The market’s payment schedule is part of the instrument. Put it in the ledger, settlement by settlement, and the backtest can show what the position actually carried.
← All posts


