How To Create a Delta Neutral Strategy
A funding rate strategy earns income from the mechanics of perpetual futures markets rather than from price direction.
The concept is simple enough to explain in a paragraph.
Running it reliably is a different problem, and it is mostly an execution challenge which is why almost every serious implementation is automated.
This article covers how the strategy works, how capital is actually allocated across the two legs, why manual execution consistently underperforms, and the part rarely written down: what breaks once real capital is running through it.
What a funding rate strategy actually is
Perpetual futures have no expiry date. That creates a problem: without a settlement date forcing convergence, nothing mechanically ties the contract price to the price of the underlying asset.
Exchanges solve this with a funding rate, a periodic payment exchanged directly between long and short position holders.
When the perpetual trades above spot, funding is positive and longs pay shorts. When it trades below, funding is negative and shorts pay longs.The payment pulls the contract back toward spot by making the crowded side progressively more expensive to hold.
The strategy follows from that.
Hold spot long and an equal-size perpetual short at the same time, and directional exposure cancels: if the asset falls, the spot position loses and the short gains by roughly the same amount. What remains is the funding payment. Net price exposure is approximately zero, and the funding rate becomes the return.
This is what's known as being "delta-neutral": You're removed from any exposure to price action.
Constructed correctly, the position is close to indifferent to whether the market goes up or down. That property is what makes the income structural: it derives from persistent demand for leverage in perpetual markets, not from a directional view.
One detail matters more than most explanations suggest: funding is paid at discrete intervals, based on the position held at that exact timestamp. Centralised venues typically settle every eight hours; some decentralised venues settle hourly.
How capital is actually allocated
Most explanations stop at "hold equal size on both legs" and move on. While that's technically true, it leaves out the part that determines how much yield a given amount of capital produces, and for how much somebody can actually run manually a strategy like this.
The naive approach is a 50/50 split: half the capital buys spot, half margins the short perpetual, both unleveraged. It is delta-neutral and it works, but it is also capital-inefficient, because the perpetual leg does not need anywhere near half the capital to hold its side of the hedge.
A more efficient allocation uses leverage on the perpetual leg only:
- 2/3 of capital → long spot, unleveraged (1×)
- 1/3 of capital → short perpetual at 3× leverage
Check the notional: 2/3 × 1 = 1 unit of long spot exposure. 1/3 × 3 = 1 unit of short perpetual exposure. The legs match exactly, so the position remains fully hedged, but only a third of the capital sits on the margin side.
The consequence is that funding capture per unit of deployed capital increases substantially versus the naive split, because the perpetual position (the leg that actually receives funding), is sized against the full notional rather than against half the capital.
The trade-off is margin headroom.
Leverage on the short leg brings the liquidation price closer, which is exactly the risk discussed further down. But 3× is conservative enough to survive ordinary volatility while still meaningfully improving capital efficiency.
Higher leverage would capture more funding per unit of capital and would liquidate sooner.
This measure of course must be adapted depending upon each asset volatility profile. Higher volatility assets might require for greater margin buffers limiting the short leveraged leg to lower amounts of leverage and maybe, no leverage at all.
Why manual execution fails
The theory says open two positions of equal size. Execution says click a buy order on one venue, then click a sell order on another, seconds apart.
Anyone can do this. The challenge remains on what happens after the orders are executed.
Between the first fill and the second, the price moves. If the spot leg fills and the perpetual leg has not, the position is briefly and entirely directional: long the asset with no hedge, carrying exactly the risk the strategy exists to eliminate.
In calm markets this is a rounding error. In the volatile conditions that produce the widest funding spreads, and therefore the most attractive entries, it can cost more than the trade was going to earn.
Slippage compounds the problem. Every leg pays a spread and a fee. Funding spreads are frequently measured in basis points, so execution quality is not a detail at the margin, it often determines whether a position is profitable at all.
Two legs executed carelessly can consume the entire expected return before the first funding payment arrives.
Then there is coverage.
Funding rates change continuously and independently across venues, and the attractive spreads appear without warning. A person cannot watch multiple venues around the clock, which means manual operators systematically capture worse rates than were actually available, not through poor judgement, but because the good windows opened while they were asleep.
This means the execution venue of preference must be carefully choosen to allow traders to access deep enough liquidity and lasting longer positive funding rates to allow for the capture of the highest amount of funding rate as possible versus, trying to capture funding rates from the shiniest venue all at once.
How automation closes the gap
Automating this strategy is not simply "the same thing, faster".
A working system has to handle several distinct jobs.
Simultaneous execution. Both legs are submitted together rather than sequentially, collapsing the window where the position is unhedged.
Continuous cross-venue monitoring. The system tracks funding rates across every venue it can trade on. Opportunity is defined by the spread available anywhere, not the spread visible on whichever tab happens to be open.
Position sizing and margin management. Leg sizes must match in notional terms, and the leveraged short requires margin maintained at a level that survives adverse moves. Sizing is a continuous constraint, not a one-time decision at entry.
Programmatic rebalancing. As price moves, the notional value of the two legs diverges and the position drifts away from neutral. Restoring neutrality requires periodic adjustment.
Fee-aware pre-trade checks. This is where naive implementations quietly fail. A funding spread that looks attractive gross can be negative once round-trip fees on both legs are included. A system that enters whenever it sees a positive spread will lose money consistently while appearing to do exactly what it was designed to do.
The correct behaviour is frequently to decline the trade. This is why execution is the moat when it comes to executing this type of strategy.
That last point separates production systems from tutorials. Most published implementations of this strategy, including several well-regarded open-source ones, treat the spread as the return. It is not.
The return is the spread net of every cost required to capture it, and the difference between those two numbers is where the strategy is won or lost.
What breaks in production
Leg risk on entry and exit
Simultaneous submission reduces the unhedged window; it does not eliminate it. Orders can partially fill. One venue can be slower than another. Under stress, an order can be rejected outright while its counterpart fills normally, leaving a live directional position nobody intended. Systems need to detect partial or failed fills within seconds and either complete or unwind the position, rather than assuming both legs landed.
Funding flipping negative mid-position
Funding is not fixed. A position opened while funding is comfortably positive can see that rate compress and turn negative as sentiment shifts, at which point the short leg pays instead of receives, and the position runs at a cost.
The decision is genuinely difficult. Unwinding immediately means paying exit costs on both legs and forfeiting the position if funding recovers within hours, which it often does. Holding through means accepting a known negative carry of unknown duration.
Neither is right in all cases, which is why this requires an explicit policy rather than an instinct.
Liquidation on the short leg
This is the failure most often missed, and the most damaging, and the leveraged allocation described earlier makes it more relevant, not less.
Being delta-neutral in aggregate does not protect the individual legs. The perpetual short is a margined position with its own liquidation price. If spot collateral sits on one venue and the perpetual short on another, a sharp upward move can liquidate the short leg while the offsetting spot gain sits somewhere the exchange cannot see and will not credit.
The position was neutral but the venue did not know that.
You are left holding spot with no hedge, having realised a loss on the leg that was supposed to be protecting you, and typically at the worst possible moment, since violent upward moves are exactly when funding spreads are widest and positions are largest.
Delta drift and rebalancing cost
A position that starts perfectly matched does not stay matched. As price moves, the notional values of the two legs diverge and residual directional exposure accumulates.
Rebalancing fixes this and costs fees every time. Rebalance too often and fees consume the funding income; too rarely and you are carrying directional risk in a strategy sold as market-neutral.
There is no universally correct threshold: it depends on volatility, fee tier, and position size, and it needs to adapt.
Venue outages during volatility
Exchanges degrade under load. APIs rate-limit, order submission slows, withdrawals pause, interfaces go down entirely. This happens disproportionately during high volatility — precisely when positions most need adjusting.
A system that assumes venue availability will eventually attempt an adjustment it cannot execute.
Counterparty and custody risk
Running this strategy means capital sitting on trading venues, exposed to the operational and solvency risk of every venue it touches regardless of how well the strategy performs. This risk is uncorrelated with market direction and is not hedged by delta neutrality. It is all a separate exposure managed through venue selection, per-venue position limits, and how much capital rests anywhere at all.
Fee drag when funding compresses
Funding rates cycle. In quiet markets they compress toward zero across most venues at once, and the arithmetic inverts: round-trip fees exceed the funding available to collect.
The correct response is to hold no position, operationally harder than it sounds, because a system built to find opportunities needs an explicit, well-calibrated definition of "not worth it" or it will trade steadily into a loss during exactly the periods when doing nothing was profitable.
Cross-venue margin transfer latency
When one venue needs margin urgently, moving collateral takes time. Withdrawals require confirmations, and networks congest during volatility. The transfer that would have prevented a liquidation may not arrive in time.
This makes margin buffers a design decision rather than an afterthought: capital held as buffer is capital not earning funding, so the trade-off between efficiency and survivability is direct and permanent.
Venue selection and cross-venue spreads
Funding rates are not uniform. The same asset can carry meaningfully different funding on different venues at the same moment, because rates reflect local supply and demand for leverage rather than a global reference.
Venue user bases differ. Some skew retail and directionally long, producing persistently positive funding. Others attract more sophisticated flow and price more efficiently. Decentralised perpetual venues often display different funding dynamics from centralised ones, both because of different participants and because some settle funding on shorter intervals.
Asset selection matters as much as venue selection, and the two interact. Comparing funding and volatility across major perpetual markets produces a non-obvious result: the assets with the deepest liquidity do not necessarily offer the best risk-adjusted funding. In our own comparative analysis across BTC, ETH and HYPE, HYPE funding rates ran consistently two to three times higher than BTC and ETH, while HYPE volatility stayed comparatively balanced, without the extreme spikes the larger assets exhibit. That combination, high funding with manageable volatility, is what actually matters for a basis position, and it is not what most operators default to.
Whether a venue is worth trading comes down to three properties, and headline funding rate is not one of them:
- Depth: can positions be opened and closed at size without moving the price against you?
- Funding reliability: does the venue calculate and settle funding predictably, including under stress?
- Withdrawal reliability: can capital actually be removed when needed? A venue with excellent rates and unreliable withdrawals is not a good venue at any spread.
Manual, self-hosted, or managed
There are three honest ways to run this strategy, and the right answer depends on what you are optimising for.
Manual execution is workable for learning the mechanics and for small size. It does not scale, the leg-timing and coverage problems above are structural, not a matter of being more careful.
A self-hosted bot suits people who want full control and can maintain infrastructure. You choose venues, set your own risk parameters, keep custody of your keys, and adapt the system as you learn. Several capable open-source implementations exist.
What they typically do not include is the operational layer described above: partial-fill recovery, negative-funding policy, adaptive rebalancing thresholds, venue-failure handling, margin buffer management. Those are not features you add in an afternoon; they are what you build over successive market cycles, usually after each one teaches you something expensive.
If you want to run infrastructure and are prepared for that learning curve, self-hosting is a legitimate choice and the control is real.
A managed agent suits people who want the exposure without operating the system. The trade-off is explicit: you accept somebody else's execution decisions and risk parameters in exchange for not building, monitoring, or maintaining any of it.
That trade-off is only worth making if the operator has genuinely solved the problems above, which mostly means having encountered them. This is what Deploy has been doing since 2018, across multiple volatility cycles, with onchain code audited by Halborn and QuillAudits. Income: Funding Rates is the delta-neutral agent that runs exactly the strategy described on this page: spot long, perpetual short, funding captured continuously, with the execution and risk layer handled.
Anyone comfortable running their own infrastructure should probably run their own. The managed route exists for everyone else, and for people who have run their own long enough to know what the operational layer actually costs to maintain.
Frequently asked questions
What is a funding rate?
A periodic payment exchanged between long and short holders of a perpetual futures contract, used to keep the contract price aligned with spot. When the perpetual trades above spot, longs pay shorts; when below, shorts pay longs. Most centralised venues settle every eight hours.
Is funding rate arbitrage risk-free?
No. It removes directional price risk, not risk in general. Funding can turn negative and cost you. The short leg can be liquidated independently even when the overall position is neutral. Venues can fail, freeze withdrawals, or become insolvent. Fees can exceed funding income in quiet markets. It is a strategy with a different risk profile from directional trading, not an absence of risk.
What happens when funding goes negative?
The cash flow reverses and the short leg pays instead of receiving, so the position carries a cost. Options are to unwind and pay exit costs, or hold and accept negative carry until funding recovers. Which is correct depends on how long the condition is likely to persist and what exiting and re-entering would cost, it requires a defined policy rather than a case-by-case judgement call.
Why use leverage on the perpetual leg if the strategy is delta-neutral?
Leverage on the short leg is what makes the allocation capital-efficient. Committing 1/3 of capital at 3× produces the same notional short as committing all of it at 1×, freeing the remaining 2/3 to hold the spot leg. Notional stays matched, so the position remains neutral, but funding capture per unit of deployed capital improves. The cost is reduced margin headroom before liquidation, which is why the leverage figure is a risk decision rather than an optimisation.
Can funding rate arbitrage be fully automated?
Execution can be. Monitoring, entry, sizing, rebalancing and exit are all mechanical and better handled by software than by a person. What still requires human judgement is policy: which venues to trust, how much capital to expose to any one of them, what buffers to hold, and when conditions warrant standing down entirely.
What capital is required?
There is no universal minimum, but fee structures create a practical floor. Funding spreads are typically small in percentage terms, so at low capital the fixed costs of execution consume the return. Capital must also be split across venues and legs, with margin buffers held back rather than deployed, meaning not all of it is earning at any given time.