The real cost of running HyperMirror: fees, slippage, funding and opportunity cost

Why a single fee number understates cost

Most automated trading products advertise one number: a subscription, a performance fee, or a percentage of volume. Ours is 0.1% of mirrored notional volume. Quoting that alone would be technically accurate and practically misleading, because in a leveraged perpetual futures system the fee we charge is frequently not the largest cost of running it.

The reason is that almost every cost in this system scales with turnover, not with the size of your account. Two users with identical capital, running for identical periods, can pay very different total costs purely because one basket's leaders traded four times as often. Cost is a function of activity, and activity is a property of the leaders, not of the fee schedule.

This note walks through every cost layer we are aware of, including the ones that are not ours and the ones that never appear as a charge at all.

Layer 1 — the builder fee: 0.1% of mirrored volume

The builder fee is Hyperliquid's native mechanism for a third party to be compensated on order flow they route. It is applied at the protocol level to the notional volume of mirrored orders, and it is capped by a maximum rate that you approve in your wallet before any trade happens. We cannot charge above that cap, and you can revoke it.

What it applies to: notional volume of orders placed by the mirroring agent in your sub-accounts. What it does not apply to: your deposits, your withdrawals, your own manual trades, your unrealised gains, your account balance, or time. There is no subscription, no management fee, no performance fee and no lock-up.

The practical consequence of a volume-based fee is worth being explicit about: it is charged whether the mirrored trade wins or loses. A volume fee aligns our revenue with activity, not with your outcome. We think a capped, protocol-enforced, upfront-visible rate is the most honest available structure on this venue, but it is not a performance-aligned one, and pretending otherwise would be dishonest.

Layer 2 — Hyperliquid exchange fees

Separate from anything we charge, the venue charges its own taker and maker fees on each fill, with rates that depend on your own volume tier. These accrue to Hyperliquid, not to us, and they apply to mirrored fills exactly as they would to trades you placed yourself.

Mirrored orders skew toward taking liquidity rather than providing it, because the objective is to follow a leader's position promptly rather than to wait for a better queue position. Passive orders would reduce this cost and increase tracking error; that trade-off is real and resolves in favour of tracking.

Layer 3 — slippage and spread

This is the cost of mirroring rather than originating. The leader's fill is observed and then reproduced, which means your order arrives after theirs, into a book their order has already moved. You cross a spread they may not have crossed, at a price that has already reacted.

It is invisible on any statement. It shows up only as a persistent difference between your sleeve's result and the leader's, which is exactly why it gets attributed to the wrong thing. In liquid majors it is usually small. In thin alt perps, during volatility, or on a position large relative to book depth, it can exceed every explicit fee combined.

Slippage scales with turnover and with the illiquidity of the markets your leaders trade, which is another way of saying it is a property of leader selection.

Layer 4 — funding: a cost or a credit

Perpetual futures pay periodic funding between longs and shorts. Holding a position on the paying side is a continuous drag proportional to position size and holding period. Holding the receiving side is income.

Treating funding as a pure cost is a common and expensive error in cost accounting. In a basket that holds both directions across sleeves, funding partially offsets, and in some regimes a carry-oriented leader is generating most of their return from it. The correct treatment is as a signed term that depends on side, size and regime.

What is reliable is that funding scales with holding period rather than with turnover, which makes it the one major cost that behaves in the opposite direction from all the others. Low-turnover leaders pay less in fees and slippage and more in funding exposure.

Cost with no line item: replacement and rebalance turnover

Every leader handover pays turnover twice: once to close the outgoing sleeve, once to open the incoming one. Every rebalance moves capital between sub-accounts and adjusts positions, which is turnover again.

These are the costs of the system's adaptivity, and they are the direct price of the features that make rule-based replacement worth having. A system that never replaces a leader pays none of this and holds decayed selections forever. There is no configuration that avoids both.

Rebalancing is user-initiated for exactly this reason: you decide when the drift is worth paying to correct.

Cost with no line item: idle capital

Capital in a sleeve that is flat — after an unwind, before a replacement leader qualifies, or while a leader simply is not in the market — earns nothing and is still committed. The same applies to a sleeve that failed to fund because of a readiness blocker.

This is a real drag on the return of the whole account, and it is one that a per-sleeve performance view hides completely, because a flat sleeve shows a flat result rather than a negative one. Measure it at the account level or you will not see it at all.

Cost with no line item: opportunity cost of Perps margin

USDC held as margin in your Perps balance is not deployed anywhere else. Whatever yield that capital could have earned elsewhere is forgone for as long as it backs the sleeves, and margin has to be sized for the worst moment rather than the average one, so some of it is idle by design.

This is not a charge and nobody collects it, but it belongs in an honest total, and it is the reason that running a basket with far more margin than the sleeves need is quietly expensive.

What drives total cost most: leader turnover

Add the layers up and the pattern is clear. Three of the four explicit layers — builder fee, exchange fees, slippage — scale with mirrored volume. Two of the three hidden costs — replacement and rebalance turnover — do as well. Only funding and idle capital scale with time instead.

That means the dominant determinant of your total cost is how much notional volume your basket generates, which is a property of the leaders in it. A basket of high-frequency leaders can cost several times more to run than a basket of position traders with identical capital, identical fee rates and identical account value.

The corollary for evaluating any copy trading product, including this one: comparing fee rates between products tells you very little. Comparing cost per unit of turnover, and then estimating turnover, tells you almost everything.

How to estimate your own cost

The arithmetic is straightforward and worth doing with your own numbers rather than ours. The structure is: sum the notional volume of your mirrored fills over a period, multiply by 0.1% for the builder fee, add your Hyperliquid fee tier's rate on the same volume, add an assumption for average slippage per fill, then add or subtract net funding from your position history.

Any worked example we could publish would be arithmetic on assumed inputs — an assumed turnover rate, an assumed slippage figure, an assumed funding regime — and would tell you nothing about what your account will actually pay. It would also invite being read as an implied return, which it is not. Your own trade feed contains the real inputs; use those.

• Builder fee: mirrored notional volume x 0.1%.
• Exchange fees: same volume x your Hyperliquid fee tier.
• Slippage: an assumption per fill; scale by fill count, not by capital.
• Funding: signed, from your own position history.
• Idle capital: share of margin not backing an active position, over time.

What we do not charge

For completeness, the absent items are as informative as the present ones.

• No subscription or monthly fee.
• No performance fee or high-water mark.
• No management fee on assets, because we do not hold assets.
• No deposit or withdrawal fee; funds never leave your own account.
• No lock-up, notice period or redemption schedule.
• No fee on your own manual trades, only on mirrored volume.

Conclusion: judge cost per unit of turnover

The honest summary is that 0.1% of mirrored volume is the visible part of a cost stack whose largest components are frequently the exchange's fees, the spread you cross to follow someone else's fill, and the turnover that adaptivity requires. Funding can cut either way. Idle capital and margin opportunity cost are real and easy to overlook.

None of that makes the system expensive or cheap in the abstract. It makes cost a function of activity, which means the right question before starting is not 'what is the fee' but 'how much will this basket trade'.

The fee you approve is visible and capped during onboarding, before any trade is mirrored, and the builder-fee note covers exactly what it applies to and what it does not.