Futures Basis vs Perpetual Funding: Two Different Costs

Two instruments, two clocks

A dated future expires. On the expiry date it settles against a reference price, and the difference between the futures price and spot must disappear. That convergence is not a market convention; it is a contractual term.

A perpetual future does not expire. To keep its price near spot, the contract uses a funding mechanism: at fixed intervals, one side of the market pays the other. The payment is not a fee to the exchange; it is a transfer between traders.

Both mechanisms charge for time, leverage and the use of someone else’s capital. They are not interchangeable, and the numbers they produce are not directly comparable without adjustment.

Basis: the price of a dated future

Basis is the difference between the futures price and the spot price. The definition sounds trivial until the sign convention changes.

In CME’s FX education material, basis is quoted as futures minus spot, driven by the interest rate differential and converging to zero at expiration. In CME’s Treasury material, the basis of a deliverable bond is quoted against the futures price adjusted by a conversion factor, which is a different construction. Agricultural basis is usually quoted as cash minus futures. The same word covers at least three conventions.

For a crypto dated future, the simplest and most common convention is:

basis = futures price - spot price

A positive basis is called contango; a negative basis is backwardation. An annualised basis converts the raw difference into a rate:

annualised basis = (futures / spot - 1) × (365 / days to expiry)

That annualisation assumes the basis is a cost of carry, comparable to an interest rate. For a cash-and-carry position, it roughly is: buy spot, sell the future, hold to expiry, and the return is the basis.

Convergence and rollover

A dated future’s basis must go to zero at expiry only if the contract settles against the same spot reference the trader is using. In practice the settlement price is defined by the exchange, and for physically delivered contracts the delivery process introduces its own costs. For cash-settled crypto futures, the settlement index is the reference.

Before expiry, the basis can widen. A trader who is long the future and short spot is exposed to that widening until convergence. If the position is leveraged, a widening basis can trigger a margin call even though the trade is “correct” at expiry.

Rolling a position from one expiry to the next replaces one basis with another. If the curve is in contango, the roll sells a cheaper future and buys a more expensive one, which is a cost. If the curve is in backwardation, the roll can be a credit. A strategy’s return is the spot move plus or minus the roll, not the spot move alone.

Perpetual funding: the periodic payment

Binance’s documentation states the mechanism directly. The funding rate is calculated from the average premium index and an interest rate component:

Funding Rate = [Average Premium Index + clamp(interest rate - premium index, 0.05%, -0.05%)] ÷ (8 ÷ N)

where N is the funding interval. The interest rate is fixed at 0.03% per day by default, which is 0.01% per eight-hour interval. The funding amount is the mark price multiplied by contract size, multiplied by the funding rate.

The direction depends on the sign:

• Positive funding rate: longs pay shorts.
• Negative funding rate: shorts pay longs.

Only positions open at the funding timestamp are charged. Binance’s default schedule is 00:00, 08:00 and 16:00 UTC, and the exchange may change the interval during extreme volatility.

The key property is that funding is not fixed at entry. It is recalculated every interval from a premium index that reflects the perpetual’s price relative to spot. A position held for a week pays whatever the rate was at each of the 21 funding times, and that path can include sign changes.

A worked comparison

Suppose bitcoin spot is 60,000. A 90-day future trades at 61,200, and a perpetual trades at 60,000 with a funding rate of 0.01% per eight hours.

Dated future. The basis is 1,200, or 2.00% of spot. Annualised simply: 2.00% × 365 / 90 = 8.11%. Compounded, (1.02)^(365/90) - 1 = 8.36%. A cash-and-carry trade that buys spot and sells the future locks in roughly that return if held to expiry, before fees and financing.

Perpetual. At 0.01% per interval and three intervals per day, a long pays 0.03% of notional per day. On one bitcoin of notional at 60,000, that is 18 per day. Over 90 days at a constant rate, the simple total is 2.70%, or 1,620; compounded, about 2.74%. Annualised simply, 10.95%; compounded, about 11.57%.

The perpetual looks more expensive in this example, but the comparison is not like for like. The future’s 8.11% is a return on capital committed to a cash-and-carry position, while the perpetual’s 10.95% is a cost paid by a leveraged directional trader who has not posted the full notional. The relevant comparison depends on the position, not on the headline rate.

The comparison also assumes the funding rate stays at 0.01%. It does not. A funding rate of 0.05% per interval for a week, followed by a negative rate for a week, produces a total that no single annualisation predicts.

What each cost actually is

The table’s last row is the one traders most often miss. Exiting a dated future before expiry does not return the entry basis; it realises whatever the basis is at exit. Exiting a perpetual stops future funding but does not refund past funding.

Why the confusion is expensive

Three specific errors follow from treating the two as the same cost:

1. Annualising one funding interval. Multiplying 0.01% by three per day and 365 days gives 10.95%, which looks like a yield. It is only a yield if the rate is constant, and funding is designed to move.
2. Comparing a funding cost with a basis return. The basis is earned on a fully funded cash-and-carry position. Funding is paid on a leveraged notional. The denominators differ.
3. Ignoring mark price. Funding is charged on mark-price notional, which can differ from the last traded price. A position’s funding cost is not simply its entry notional times the rate.

What to check before choosing

• Holding period. A dated future with a known expiry suits a position with a known horizon; a perpetual suits a position that may need to exit at any time.
• Curve shape. Contango makes rolling a long future expensive; backwardation makes it cheap or profitable.
• Funding history. Look at the distribution, not the current rate. A rate that is usually positive but occasionally spikes is a different risk from a stable one.
• Liquidity at the expiry. A dated future’s liquidity migrates to the next contract as expiry approaches, which raises the cost of exiting.
• Margin and liquidation. Basis widening can liquidate a leveraged cash-and-carry trade before convergence; funding spikes can liquidate a perpetual position.

The Bybit funding fees worked example walks through the payment arithmetic on a specific venue, and the OKX funding rate, mark price and open interest guide covers how the premium index is built. Both are useful for the perpetual side. For the futures side, the gold ETFs vs gold futures comparison shows the same cost-of-carry logic in a market with no funding mechanism at all.

The short version

Basis is a price difference that converges because a contract expires. Funding is a cash flow that repeats because a contract does not. One is locked in at entry and realised at exit; the other is repriced continuously and paid as it goes. They both compensate someone for carrying a position, but they do it on different clocks, and the numbers only compare after the holding period and the capital base are made explicit.