We will never tell you what you'll earn. We'll show you exactly what we control.
You don't get to control the market — you get to control the machine that touches it.
HyperCore is non-custodial copy-trading infrastructure: a Rust risk and execution core plus a multi-tenant control plane you white-label and run for your own customers' capital. Each account runs a portfolio of copy bots — many leaders, independent caps, one kill switch — and one login manages many accounts as separate books. The platform can read, configure, and kill — it never holds withdrawal rights and never sees a key. Not hope. Not returns. Control.
replayed leader equity · R1 dataset · drawdown survived, kill not triggered
Most strategies lose. We show you which — and by how much.
A returns promise is the tell. We don't make one. Our strategy lab replays a real leader against our strategies, and most of those strategies lose on paper — before a cent is risked — and the product surfaces the losers in the same plain numbers as the winners. At ~$3,000 capital with ~9 bps round-trip cost, the prior for any strategy is that it loses. We measure that, and we show it.
strategies falsified on the first replay — verdict still pending the R2 re-record. That is the lab working.
Strategy-lab scoreboard
replay| Strategy | Tape | Realized PnL | Verdict |
|---|---|---|---|
| Mirror | 148 maker fills | −$12.08 | FALSIFIED |
| Fade | 27 taker fills | −$0.18 | FALSIFIED |
| Maker | thin tape · <30 samples | −$0.0154 | INCONCLUSIVE |
Mirror blocked 38.8k of its own orders at the risk gate — confirming the “avoid” finding. Maker's tape is too thin to judge; R2 re-records with a full trade subscription before any verdict is final.
This is what an operator sees at 3am.
One live account in full: exposure, the risk engine and its breakers, fills, leader scores, order history — every cell a live feed. (The fleet view rolls every account up on one screen; this is one book up close.) Calm until something moves; then the one thing that changed is the only thing that's loud. The session on screen is a real losing day, −$26.45, and the instrument shows it plainly. We don't hide that — it's built to make it legible.
−$26.45 — a real losing session, shown in plain numbers — never massaged, never hidden.
The control bar — flatten, cancel, halt, resume — the kill switch, quiet at rest, token-gated and audit-logged.
54 live mirrors — every mirrored order and fill as one row — the Bloomberg-style density an operator actually reads.
- See the operator console →
Your customers get the calm sibling. Same engine, same truth.
One north-star number — their copy P&L — first. Then the copy bots they run (here, two leaders diversified under one account), their own kill switch, and a plain read on how their money is doing. They switch between their books from the top bar and never see a key. No confetti. One system, two densities: Bloomberg for the desk, a calm statement for the customer — under the operator's own brand.
+$5,651.59 — one north-star — their copy P&L — with the same living trace, settled and calm.
2 copy bots — Atlas and Meridian in one account — two leaders, each its own ratio and cap; “Aurora Desk” up top switches between their books.
Stop copying & flatten — the customer's own kill switch — they are never locked into a position.
Aster Capital, powered by HyperCore — white-label chrome: the operator's brand on top, the same honest engine beneath.
- See the trader dashboard →
We can read it, configure it, and kill it. We can never withdraw it.
Trade-only keys. On connect we rehearse a withdrawal and confirm the network rejects it — then show you that rejection as your proof. Non-custodial isn't a feature. It's the whole boundary.
- Configure your copy settings
- Read your positions & PnL
- Kill — stop & flatten
- Withdraw your funds
- Move your funds
- See your private key
- 1Connect a trade-only key
- 2We rehearse a real withdrawal
- 3The network rejects it
A rehearsal at connect time — the same check the trader dashboard shows as “Withdraw rehearse passed”.
One touch, confirmed, audit-logged — the operator's, and every customer's, dead-man's switch.
Two dials of scale.
Copy trading is rarely one leader and one account. HyperCore scales two independent ways — and never touches custody on either axis. A copy bot follows one leader; an account runs many bots; a login runs many accounts.
Run a portfolio of copy bots
A copy bot follows one leader at a ratio you set, with its own exposure cap. Stack several in one account to diversify across leaders — one account-level kill switch and dead-man's switch cover every bot at once.
One login, separate books
Keep your prop desk, a community fund, and your own capital as fully isolated accounts — separate risk limits, separate billing, separate trade-only keys. Switch between them in a click; nothing bleeds across.
Copying is only honest if the copy tracks. Here is the gap we measure.
We measure slippage between the leader and the copied account on every fill — and we show the gap. Tight tracking is the difference between copy-trading and a brochure.
- Median tracking error
- 4.2 bps
- Worst case
- 11.8 bps
- Fills tracked
- 1,284
Illustrative replay — example figures, not a shipped measurement.
Run HyperCore as your product.
An isolated instance per operator: your brand on the chrome, your vetted-leader catalog, per-tenant risk limits, every account's copy bots and exposure on one screen, and a per-tenant kill switch plus dead-man's switch you control. A prop desk, a community lead, or a small fund — one coherent product, white-labeled. Non-custodial end to end; per-tenant breakers, KYC geo-block, audit trail.
Talk to us about white-labelingA Rust core that can be killed.
This is real engineering, not a wrapper. The frontend only makes the safety legible — the engine is what's in control.
Hyperliquid validators sit in one region — we don't pretend to beat physics, and we say so. Honest latency beats a fake number.
- Exact-decimal money math
- rust_decimal everywhere — f64 on a price is a bug, not a rounding error.
- Three independent breakers
- exposure cap, margin utilization, and a daily-loss limit — each can halt alone.
- A dead-man's switch
- always visible; if the heartbeat stops, copying stops.
- Testable before a cent moves
- every behavior runs against a mock exchange; testnet before mainnet, always.
- Drilled in live validation
- kill-9 and flatten drills passed against the running engine.
HyperCore is infrastructure, not advice.
We do not promise, project, or guarantee returns. Copy trading can lose money, including all of it. Most strategies lose — we built the lab to show you that, and the engine to limit the damage. This is not investment advice.