What the basis actually is.
Most perpetual futures venues run a funding-rate mechanism. Every eight hours, or every hour depending on the venue, longs and shorts settle a small flow against one another so the perpetual price stays anchored to the spot. When the perp trades above spot, longs pay shorts. When the perp trades below, shorts pay longs.
A long-spot, short-perp pair of equal notional has no directional exposure on paper. Its P&L from price moves cancels. What remains is the funding flow, payable to whichever side of the market is being paid by the venue's mechanism at that moment.
Worked example
0.01%/ 8h ≈ 10.95%annualised
At a steady 0.01% per eight-hour interval paid to the short side, a fully hedged spot–perp position earns roughly 10.95% annualised before any friction is taken out — execution, the cost of borrowing or holding the spot leg, and the haircut on whatever collateral the perp sits against.
Friction matters and we'll get to it. The shape of the income is what's worth understanding first.
Why we call it the cleanest carry available.
Three structural properties.
Paid by protocol, not by counterparty. The funding flow is a mechanical settlement embedded in the venue's risk system. It does not require a market-maker to honour a two-sided quote, or a swap counterparty to perform under stress. The credit risk is the venue itself, which is a different problem with a different mitigation — but it is the same problem you already underwrite by trading on the venue. There is no additional carry-specific counterparty bridge.
Direction-agnostic when the hedge is held. A spot long and a perp short of equal notional move together against price. A 5% rally on the spot is a 5% loss on the perp. What remains is the funding settlement. This is not theoretical — it is mechanical. It is also not free; the hedge must actually be sized and rebalanced correctly, which is the entire second half of this note.
Observable. Funding rates are public. They are broadcast by the venue, recorded on chain when settlement is on chain, and visible via API everywhere else. You do not need to wait for an attribution statement to know what you earned. We treat this as a precondition: a return stream that is observable as it accrues sits in a different operational category from one that has to be reconstructed at month-end.
The discipline required to actually collect it.
The carry is small per unit. Ten to thirty basis points per day in normal regimes; less in benign markets; more during stress. Holding that yield against operational reality requires three things.
The hedge has to be exact. A 1% drift in the hedge ratio against a sharp directional move can wipe out a week of carry. We measure hedge drift continuously and pull positions back into balance within a tight band, both on individual instruments and at the book level. Rebalancing has its own cost; the discipline is to net it against the carry we are actually capturing, and to refuse positions where the realistic rebalancing footprint eats the income.
The position has to be sized to liquidity. Carry trades blow up not when the trade is wrong, but when the unwind is. A position sized to the average book depth at a calm moment is a different position from one sized to the depth available when something has broken at a venue. We size to the second case.
There has to be a gate for structurally negative funding. When funding flips sign — extended bear regimes, certain altcoin perps after a liquidation wave, the entire perp complex during a sustained funding-arbitrage rotation — the income reverses. The discipline is not to predict these regimes; it is to set drawdown thresholds that automatically de-risk when realised funding turns negative against the book, and to refuse to scale into venues whose funding is structurally extreme as a yield play.
Where it sits in our attribution.
We attribute the return of the book across spreads, funding and basis, pool fees, and inventory carry. The proportions vary by month; an illustrative composition is roughly:
38%Spreads
30%Funding and basis
20%Pool fees
12%Inventory
Funding and basis is rarely the largest contributor in a single month. Its character is steady rather than heroic — small, repeated, mechanical, observable as it accrues. Over multi-month windows it tends to be the most stable line in the attribution, which is part of why we weight it the way we do.
We do not target a specific funding contribution. We target a market-neutral book with documented hedge bounds, and the funding line is what falls out when we run that book against the venue set we have approved for capacity.
When it pays nothing.
There are weeks when funding is flat or negative across the venues we'd allocate to. We treat those weeks the same way we treat regimes where any single return stream stops paying — keep the book inside policy, refuse to chase yield by overweighting venues with extreme funding (which is usually a signal of crowded positioning rather than free carry), and let the other components carry their share.
The cleanest carry is the one we can keep collecting through the regimes we did not pick.
Note on disclosure.
This is an editorial framing of a single return source within the book. Net realised contribution varies by venue, regime, and execution, and is not an indication of any specific historical or expected return. Detailed methodology, historical attribution, and the venue set under which it is collected are shared with qualified allocators under confidentiality. See Disclosures for standing notices.