Hyperliquid Copy Trading: Does It Exist? What 9,477 Vaults Show (2026)

Search for Hyperliquid copy trading and you will find guides, comparison posts and a dozen apps that promise to mirror a whale's perp positions into your account. Search Hyperliquid's own documentation for the same phrase and you will find nothing. We downloaded the entire documentation bundle on 11 October 2026 — 526,525 bytes, every page the protocol publishes — and the string "copy trading" appears zero times. So does "copytrading", so does "social trading". The 159-entry documentation index does not contain the word "copy" at all.

That is not a gotcha. It is the single most useful fact about the topic, because it tells you that everything sold as Hyperliquid copy trading is either a repurposing of a different feature or a third-party product wrapped around the public API. This article is about what those two things actually are, and what the on-chain data says about whether either one works.

Everything below comes from two public Hyperliquid endpoints that anyone can re-run: the full vault registry and the full trader leaderboard. We pulled both on 11 October 2026. That is 9,477 vaults and 47,237 ranked accounts. No estimates, no vendor marketing, no numbers borrowed from another article.

No. Hyperliquid has no native copy trading feature. There is no button that says "copy this trader", no follower setting, and no page in the docs describing one. What exists are two separate things that people call copy trading:

• Vaults. A pooled account. You deposit USDC into a vault, a leader trades the pool, and you share the result in proportion to your deposit. This is native, on-chain and built into HyperCore. It is not mirroring — there is nothing to mirror, because you and the leader are in the same account.
• Third-party mirroring via agent wallets. An outside app watches a wallet you choose and places matching orders in your Hyperliquid account using an API wallet you approve. This is real 1:1 copying, but Hyperliquid does not build it, endorse it or document it. It is built on two general-purpose primitives — agent approval and builder codes — that the protocol provides for any application.

The distinction matters because the two have completely different risk shapes. In a vault, your money sits in someone else's account and you cannot control position size, leverage or exit. In agent-wallet mirroring, your money never leaves your own account but a third party can open and close positions in it until you revoke the key. If you are new to the category, what crypto copy trading means covers the general model before you pick a venue.

Our own Hyperliquid review covers the exchange itself — fees, funding drag, the JELLY incident. This article goes one level deeper on the copying question specifically, because the leaderboard data turns out to say something the review did not.

A vault is a shared trading account with a profit-share attached. The mechanics below are all stated in Hyperliquid's documentation as of 11 October 2026, and they are stricter than most guides let on.

Rule	What the docs say
Leader profit share	10% of profits, taken when you withdraw
Cost to create a vault	10,000 USDC fee, distributed like trading fees
Leader minimum deposit	100 USDC at creation
Leader skin in the game	Must hold at least 5% of the vault at all times
What a vault can trade	Validator-operated perps only — not spot, not HIP-3 perps
Withdrawal lock-up	1 day for user vaults, 4 days for HLP, reset on each deposit
Protocol vaults	No fees and no profit share

Three of those are worth sitting with.

The asset restriction is severe. A vault can only trade validator-operated perpetuals. It cannot touch spot markets and it cannot touch HIP-3 perps — the permissionless, builder-deployed contracts that are where a large share of Hyperliquid's newer markets live. So a leader whose edge is in spot or in a HIP-3 market literally cannot run it through a vault. That alone rules out a meaningful slice of the traders you might want to follow.

The withdrawal mechanism can force the leader out of a position. The docs describe the cascade plainly: when you withdraw and the vault does not have enough free margin, open orders are cancelled in increasing order of margin used, and if that is still not enough, 20% of positions are automatically closed, repeated until the withdrawal can be processed. Your exit is another depositor's forced liquidation of a position they wanted to hold. Leaders can opt into always closing proportionally to keep liquidation prices similar, but that is a setting, not a guarantee.

Hyperliquid itself calls this system legacy. The documentation page is titled "HyperCore vaults (legacy)" and says the system was introduced in 2023, does not support HIP-3 or spot, and that "additional features should be built permissionlessly by builders on the HyperEVM". The protocol is not improving vaults. It is pointing at the exit.

Hyperliquid publishes the complete vault registry as a single JSON file. On 11 October 2026 it contained 9,477 vaults created since February 2023. The population breaks down like this.

Measure	Value
Vaults ever created	9,477
Closed	6,387 (67.4%)
Still open	3,090
Of those, the HLP complex	8 (one parent, seven children)
Open vaults run by a human leader	3,082
Total vault TVL	$275,012,302
Held by HLP (the protocol vault)	$180,268,701 (65.5%)
Held by leader-run vaults	$94,743,601 (34.5%)
Median open leader vault	$100

One note on method, because it is the trap in this dataset. HLP is a parent vault with seven children, and the parent's reported TVL already includes the children's capital. Naively summing every open vault gives $414.7M. Subtracting the $139.7M of double-counted child capital gives the real figure of $275.0M. If you see a bigger number quoted anywhere, that is why.

HLP is the Hyperliquidity Provider — the protocol's own market-making and liquidation vault. It has no trading leader to follow, no profit share, and a strategy that is "be the house". It holds 65.5% of every dollar in the Hyperliquid vault system. When an article tells you "there is $275 million in Hyperliquid vaults", two thirds of that is a market-making fund, not someone's trading strategy you can ride.

Strip HLP out and the copy-trading surface is $94.7 million across 3,082 vaults. For scale, the top 100 traders on Hyperliquid turned over $50.1 billion in the last 30 days alone.

This is the number that reframes the whole category. The median open leader vault holds exactly $100 — which is precisely the minimum the leader is required to deposit at creation. The median vault has never received a single dollar from anyone else.

• 1,666 of 3,082 vaults (54.1%) hold $100 or less. More than half have no outside money at all.
• 334 (10.8%) hold less than one dollar.
• 2,570 (83.4%) hold less than $1,000.
• 2,857 (92.7%) hold less than $10,000.
• Only 225 hold $10,000 or more, and only 80 hold $100,000 or more.

Concentration follows from that. The single largest leader vault holds 14.3% of all leader-vault money. The top 10 hold 63.5%. The top 50 hold 90.6%. So the practical choice set is not 3,082 vaults — it is a few dozen, and the long tail is empty shells.

Liveness is worse still. Only 645 of the 3,082 open vaults (20.9%) are both funded above the leader minimum and show any profit or loss movement in the last 30 days. Four out of five open vaults are parked.

The monthly creation count is the clearest signal in the file. Hyperliquid vault creation did not slow down. It stopped.

Month	New vaults
December 2024 (peak)	1,087
January 2026	133
February 2026	159
March 2026	220
April 2026	83
May 2026	10
June 2026	5
July 2026	3
August 2026	4
September 2026	6
October 2026 (to the 11th)	1

Twenty-nine vaults have been created in the last six calendar months, against 220 in March 2026 alone. The 10,000 USDC creation fee is the obvious cause — it makes a vault a $10,000 marketing expense for a product whose median holder attracts $0 — and the documentation calling the system "legacy" is the other half of the explanation. Whatever the cause, the supply of new leaders has gone to roughly zero, which means the choice set you are picking from today is the choice set you will be picking from next year.

Here is the measurement that decides the question. Hyperliquid also publishes its full trader leaderboard — 47,237 ranked accounts, each with profit, ROI and volume over day, week, month and all-time windows. We joined it to the vault registry to ask one thing: if you rank traders by how much they actually made in the last 30 days, how many of them can you deposit alongside?

Ranked by 30-day PnL	Reachable through a vault	Share
Top 10	0	0%
Top 25	0	0%
Top 50	0	0%
Top 100	4	4.0%
Top 500	17	3.4%
Top 1,000	32	3.2%

"Reachable" is generous here: it counts both accounts that are vaults and accounts whose owner separately runs a vault somewhere. Even on that loose definition, not one of the 50 best-performing traders of the last 30 days has a vault you can deposit into. The first one appears at rank 52.

Those top 100 accounts made a combined $312,184,013 on $50,126,956,561 of volume in 30 days, and the cut-off to make the list was $1,153,554 of profit. Ninety-six percent of that cohort is simply not addressable by Hyperliquid's native mechanism. Thirteen vaults appear anywhere in the top 1,000; the best-ranked is BredoStrategy at number 52, with $6.5M of deposits.

This is the entire commercial reason the third-party copy apps exist. They are not competing with a native feature — they are routing around the absence of one. It is also why the keyword tails for this topic are full of product names rather than how-to phrasing.

Worth keeping in proportion, though: of the 18,924 accounts that traded at all in the last 30 days, 57.5% ended the period up and 42.5% ended it down, with a median 30-day ROI of 2.24%. The leaderboard is not a list of geniuses. Our broader look at whether copy trading is actually profitable applies here without modification.

Restricting to the 225 leader vaults holding $10,000 or more — the only ones where performance means anything — the reported APR distribution is:

• 80 of 225 (35.6%) are negative. Better than a coin flip, worse than the marketing.
• Median APR: 14.03%.
• Range: −82.0% to +1,859.9%.

Treat the top of that range with suspicion rather than excitement. APR on a young vault is an annualised figure extrapolated from a short window, so a leader who tripled a small book in two months reads as a four-figure percentage. Several of the highest-APR vaults in the file are under eight months old. The right way to read a vault page is max drawdown and age first, headline APR last — the same discipline we argue for in copy trading risk management.

For comparison, HLP — the protocol's own professionally run market-making vault, with $180.3M of deposits and a four-day lock-up — is reporting roughly 3.4% APR. That is the honest benchmark for what "institutional-grade and boring" looks like on this venue.

And the survivorship problem is measurable here too. Of the 100 traders with the largest 30-day profits, 13 are all-time negative. They had a good month inside a losing career. Any leaderboard you copy from — Hyperliquid's, an exchange's, or ours — ranks on a window, and a window hides that.

Every third-party Hyperliquid copy app is built on the same two documented primitives. Understanding them is how you tell a reasonable product from a dangerous one, because the mechanics are identical across all of them and only the fee and the risk controls differ.

Agent wallets (also called API wallets). You sign an approveAgent action that authorises a specific address to sign orders on your behalf. The key facts from the docs: a master account gets one unnamed agent plus up to three named ones, with two more per sub-account; a named agent can carry an expiry, and that expiry can be at most 180 days in the future. Critically, an agent can place and manage orders but cannot move funds — withdrawals still require your own wallet. This is a genuine non-custodial permission, and it is revocable at any time.

The documented footgun is nonce reuse. Hyperliquid stores the 100 highest nonces per signer, and the docs warn in bold that once an agent is deregistered its nonce state may be pruned, so previously signed actions can be replayed. The guidance is explicit: never reuse an agent wallet address. If a copy app asks you to re-approve an address you have used before, that is a real reason to walk.

Builder codes. This is how the apps get paid, and it is charged by Hyperliquid itself rather than invoiced to you. You approve a maximum fee rate for a specific builder address with an approveBuilderFee action — which must be signed by your main wallet, not by the agent — and from then on orders routed by that builder can carry a fee up to the ceiling you approved. The limits:

• Maximum 0.1% on perps and 1% on spot. That is the protocol-enforced ceiling, not a convention.
• Ten active builder approvals per user, maximum.
• A builder must hold at least 100 USDC in perps account value to collect.
• Builder codes override referral codes for the order they are attached to. If you joined through a referral link expecting a fee discount, routing through a copy app can cancel it.
• Every fill carrying a builder code is published daily as a public file at stats-data.hyperliquid.xyz/Mainnet/builder_fills/{builder}/{YYYYMMDD}.csv.lz4. You can audit what an app charged you.

That last point is unusually good disclosure and nearly nobody uses it. If a copy app tells you its fee is 0.05%, the fills file is where you check. Hyperliquid's own documentation states that independent teams have generated more than $65M in revenue through builder codes, which is a fair measure of how large the wrapper economy around this exchange has become.

Add the layers up, because no single page shows you all of them.

Cost layer	Vault route	Third-party mirroring
Base taker fee	0.045% at tier 0	0.045% at tier 0
App fee	None	Builder fee, up to 0.1% of notional
Profit share	10% to the leader, on withdrawal	Varies by app, often charged on top of the builder fee
Funding	Paid by the pool, pro-rata to you	Paid by you directly
Tracking error	None by construction	Real — you fill after the leader does
Exit friction	1-day lock-up, reset on every deposit	None — revoke the agent and close

Two things in that table deserve emphasis.

The builder fee can be more than double the exchange fee. On $10,000 of copied perp notional, the base taker fee is $4.50 and the maximum builder fee is $10.00 — 2.22 times the cost of the exchange itself. The app is the expensive part, not the venue. And unlike a profit share, it is charged on notional, so you pay it on losing copies too.

Vault volume does not count toward your fee tier. Hyperliquid's fee page states that "vault volume is treated separately from the master account". A vault therefore builds its own rolling 14-day volume from scratch and a small vault sits permanently at tier 0, paying the full 0.045% taker rate, while the same leader trading their personal account might be at tier 3 paying 0.030%. Depositors eat the difference. This is a structural drag on small vaults that no vault page shows you.

Layer the lock-up on top and the picture is clear enough: the vault route is cheaper per trade and strictly worse on control, and the mirroring route is the reverse. Neither is free, and both put 100% of the losses on you — a point we make at length in the risks of copy trading.

The two models are not variations of each other. They differ on custody, on what you are exposed to, and on what you can stop.

	Hyperliquid vault	Solana on-chain copy trading
Where your funds sit	In the leader's pooled account	In your own wallet, non-custodial
Instrument	Leveraged perpetuals	Spot tokens, no liquidation price
Funding cost	Continuous, paid to the other side	None
Who you can follow	3,082 vaults, of which ~645 are live and funded	Any wallet on a verified on-chain leaderboard
Position sizing	Fixed — you own a share of the pool	Yours to set, per trade
Getting out	1-day lock-up, may force-close positions	Revoke the copy key, sell when you like
What you pay	10% of profits plus pooled trading fees	Performance-based — uwuu charges only when you profit

uwuu sits on the right-hand column. You keep custody, you pick a trader from a verified on-chain leaderboard, a copy key signs spot trades in your own wallet with sub-400ms execution, and smart trade filtering screens what gets mirrored. There is no lock-up to wait out and no pool whose other depositors can force your position closed. The mechanics are walked through in how to copy trade on Solana, the architectural argument for keeping funds in your own wallet is in decentralized copy trading, and the wider field is surveyed in our guide to the best Solana trading bot and in the best crypto copy trading platforms.

That is a genuine trade-off, not a clean win. Perps let you go short and let you size up with leverage; spot does neither. If directional leverage is the point, Jupiter Perps and Drift are the Solana-native equivalents worth reading about, and the broader venue comparison lives in crypto leverage trading platforms. What the data in this article argues is narrower: if your reason for choosing Hyperliquid is copying someone else, the native product cannot reach 96% of the people worth copying, and the route that can reach them is an unaffiliated app charging up to twice the exchange fee.

If you are going the native route anyway, these are the checks the data suggests actually matter.

• Check the vault is alive, not just open. Four out of five open vaults show no PnL movement in 30 days. Look at the recent trade history, not the all-time chart.
• Check TVL above $100. A vault sitting at exactly the leader minimum has convinced nobody, including people who can see the same page you can.
• Read max drawdown and age before APR. A four-figure APR on a vault under a year old is an extrapolation, not a track record.
• Confirm the leader still holds their 5%. The rule is enforced, but the leader can be at exactly the floor. A leader with 5.0% has far less at stake than one with 40%.
• Plan the exit before the entry. Every new deposit resets your one-day lock-up, and a large withdrawal can trigger the 20% position-closing cascade. Do not deposit money you may want on a weekend.
• If you use a third-party app instead, approve a named agent with a short expiry, approve the smallest builder fee the app will accept, and check the published fills file afterwards. Never re-approve an agent address you have used before.

None of this removes the core risk, which is that you are trusting a stranger's judgement with leverage attached. Sizing is still the only control you fully own; position sizing and stop-loss discipline do more for outcomes here than leader selection does.

Our Hyperliquid review, published in May 2026, says the HLP vault "has historically returned 30-50% APR". Measured against the protocol's own vault endpoint on 11 October 2026, HLP is reporting approximately 3.4%. Whatever was true of earlier periods, quoting a 30-50% range today would badly mislead anyone sizing a deposit, and we were wrong to leave it standing. That review is queued for a full update.

The same review also lists third-party copy frontends by name. We would now frame that section differently: the apps in this category turn over quickly, the ones that rank in search are not necessarily the ones that still work, and the mechanism — agent approval plus a builder fee — matters far more than any particular brand. Judge them on the expiry they ask for, the builder fee ceiling they request, and whether their published fills match what they told you.

Does Hyperliquid have copy trading?

Not as a native feature. Hyperliquid's documentation, 526,525 bytes of it as of 11 October 2026, does not contain the phrase "copy trading" anywhere. The closest native product is the vault system, where you deposit into a leader's pooled account rather than mirroring their trades into your own. Actual 1:1 mirroring only exists through third-party apps built on Hyperliquid's agent-wallet and builder-code primitives.

How do Hyperliquid vaults work?

You deposit USDC into a vault and own a proportional share of it. The leader trades the pooled balance, and when you withdraw you receive your share less 10% of your profits, which goes to the leader. User vaults have a one-day lock-up that resets on every deposit; the protocol-run HLP vault has four days. Vaults can only trade validator-operated perpetuals — not spot and not HIP-3 perps.

Can I copy a specific Hyperliquid trader?

Only if that trader runs a vault, and almost none of the best ones do. Ranking the 18,924 accounts that traded in the last 30 days by profit, none of the top 50 has a vault you can deposit into, and only 32 of the top 1,000 do. To follow any other wallet you need a third-party app that places matching orders in your account through an approved agent key.

What does Hyperliquid copy trading cost?

Through a vault: the pooled trading fees, which start at 0.045% taker because vault volume does not count toward the leader's personal fee tier, plus 10% of your profits. Through a third-party app: the same base fee plus a builder fee of up to 0.1% of notional, which on $10,000 of copied volume is $10 against $4.50 of exchange fees. Many apps also add their own profit share.

Is a Hyperliquid vault safe?

Your funds sit in the leader's pooled account, so you are exposed to their positions, their leverage and their liquidations, and you cannot close a position yourself. Of the 225 vaults holding $10,000 or more, 35.6% are reporting a negative APR. The leader must keep at least 5% of the vault, which is real but modest alignment. Treat it as handing money to a stranger with leverage, because that is what it is.

Why are so few new Hyperliquid vaults being created?

Creating a vault costs a 10,000 USDC fee, and Hyperliquid's own documentation now labels HyperCore vaults "legacy" and directs builders to build vaults on HyperEVM instead. The effect is visible in the registry: 220 vaults were created in March 2026 and only 29 in the six calendar months since April.

Is copying on Solana different from copying on Hyperliquid?

Structurally, yes. Solana on-chain copy trading mirrors spot trades into a wallet you control, so there is no lock-up, no funding bill, no liquidation price, and no pooled account whose other depositors can force your position closed. The trade-off is that spot gives you no leverage and no way to go short. Which matters more depends on whether you want exposure to a trader or exposure to leverage.