The anti-churn layer for perp DEXes · Hyperliquid-native · 2 of 3 founding seats open

Your DEX isn't
acquisition-starved.
It's leaking traders
faster than you
can replace them.

Hyperliquid lost ~60% of perp market share in 30 days. Traders migrate the moment liquidation becomes the only safety net you offer. RiskGuard is the one-line SDK that keeps their accounts alive — longer lifetimes, compounding fees, no more renting users from the liquidation engine. See the 12-month math 👇.

FOUNDING PILOT · 2 OF 3 SEATS OPEN 6mo at cost · rate locked for life · no renewal obligation
🤝 90-day pilot guarantee: if your cohort's cumulative fees don't move vs baseline, you walk — we delete the agent keys the same afternoon. Our downside, not yours.
+$353K incremental Y1 fees / 1K traders
Month 3½ cumulative break-even
3.1× account lifetime extension
< 48 hrs to first risk-managed trade
Live on Hyperliquid mainnet · managing real accounts
In conversation with 8 DEX founding teams
SOC 2 engagement open · MiCA/CFTC audit trail shipping
THE CHURN PROBLEM sources verified · 2025 data
Perp DEX share · Sep→Oct 2025
−60%
Hyperliquid collapsed from ~70% to ~10% of daily perp volume in 30 days. Traders migrate at launch-day speed. Src: 21Shares
Leveraged account outcome
95%
lifetime liquidation rate. The 5% who survive are the only customers you still have next year — and they pay 8–12× more lifetime fees. Src: CCAF 2020–2023
50× leverage · median survival
< 2hrs
Most of your "daily active traders" stop being customers before lunch. You fund their acquisition anyway. Src: MEXC
Crypto trading apps · day-7
24%
Three in four users churn before their second trading session. Acquisition spend compounds; retention doesn't — unless you engineer for it. Src: Adjust
RISKGUARD · we keep the other 75% alive
Hyperliquid dYdX Drift Jupiter Perps Paradex Vertex GMX Aster Apex Orderly Perennial Lighter Aevo GRVT Hyperliquid dYdX Drift Jupiter Perps Paradex Vertex GMX Aster
§ 01 / the asymmetry

Your platform loses when a trader blows up.
They just don't come back.

Trader outcome
95.2%
of leveraged accounts are liquidated within their lifetime. The survivors are the only customers you have twelve months from now.
Src: CCAF, 2020-2023 dataset
Market momentum
26%
of global perpetual futures volume now clears on DEXes — up from 2.7% two years ago. Retention is the new battleground.
Src: CoinGecko / Cointelegraph 2025
What exists today
0
perp DEXes offer pre-trade risk enforcement to their traders. Liquidation is the only safety valve. We think that's a mistake.
Src: competitive audit Q1 2026
§ 02 / the math

$1.01M. Or $660K when you let them bleed out.

Your DEX doesn't have an acquisition problem. It has a churn problem. Same 1,000 traders, 12 months apart — the protected cohort earns you $1.01M, the unprotected one caps at $660K because 95% of them liquidate. One SDK call keeps their accounts alive and closes the $353K gap. Tier-II license already netted out.

Y1 cumulative fees
+53%
per 1,000-trader cohort vs. the unprotected baseline. +$353K in additional DEX fees over 12 months.
Break-even point
M
month the protected cohort's cumulative fees overtake the unprotected one. Every month after is pure incremental revenue.
Account lifetime
3.1×
median survival multiplier under rule-enforced risk controls vs. the liquidation-only baseline. More at-bats per trader.
12-month cumulative fees · 1,000-trader cohort
USD · $2K avg account · ~$15K daily volume / active trader · 0.06% DEX fee · liquidation revenue included in baseline
WITH RISKGUARD
WITHOUT (STATUS QUO)
$0 $200K $400K $600K $800K $1M M0 M1 M2 M3 M4 M5 M6 M7 M8 M9 M10 M11 M12 CUMULATIVE FEES (USD) MONTHS AFTER PILOT START CROSSOVER MONTH 3½ +$353K +53% Y1 $660K $1.01M
The counterargument, taken seriously

"But won't we lose liquidation revenue?"

~2.3%
of Hyperliquid's $844M 2025 revenue came from liquidation capture (HLP vault yield).
0%
of dYdX liquidation fees go to the protocol. 100% funds the insurance pool (which lost $9M in a single 2023 attack).
8–12×
more lifetime fees from a surviving trader vs. one liquidated in month 2. Trading fees are 70–85% of DEX revenue — liquidations are a rounding error on the way to the real number.

Liquidation revenue is mostly a rumor DEX founders tell themselves — the real numbers show it's ≤3% of platform revenue and often flows to LPs / insurance funds, not the protocol. Meanwhile, a trader who blows up in month 2 never pays the other 22 months of fees you just walked away from.

Argue with every number.

This is a model — not a reported figure from a customer cohort (we don't have one yet). Every input below is either sourced or labeled estimate. Each pilot re-runs the math on your actual trader data before renewal.

What changes the answer most?
Monthly survival delta is the dominant variable. If RG only extends survival from 70% to 80% monthly (not 88%), the Y1 delta falls to +21%. We show the conservative band in every pilot proposal.
Input Value Source
Cohort size 1,000 active perp traders est model unit
Average DEX fee rate 0.060% per trade src CoinGecko perp DEX avg
Daily volume / active trader ~$15,000 est $2K equity × ~7× turnover
Trading days / month 22 est std weekday count
Monthly ARPU — unprotected ~$198 derived (vol × days × fee)
Monthly ARPU — protected ~$150 (−25%) est tilt-trade filter cost
Liquidation revenue share ~2.3% of total src Hyperliquid '25 HLP / total fees
Monthly survival — unprotected 70% src derived from CCAF 95.2% lifetime liq
Monthly survival — protected 88% est prop-firm rule-enforced analogue
Annual RG cost (Tier II) −$42K subtracted from delta priced in · net shown: +$311K
LIVE ON HYPERLIQUID MAINNET · the rules firing in real time
risk_guard.py · LIVE CONNECTED · HL MAINNET
§ 03 / the five rules

Five guardrails that cancel orders before the order book sees them — not after the liquidation engine fires.

01
Daily loss kill-switch
When the trader's intraday drawdown crosses a configurable threshold, every open order cancels and new entries are blocked until the next session. Existing positions can still be closed — never flipped.
default: −5% resets: 00:00 local granularity: per-account
02
Daily profit lock-in
The asymmetric twin. When the trader is up significantly, we cancel all pending entries so they can't give it back chasing the next setup. The feature your highest-volume whales will thank you for.
default: +100% opt-in per user
03
Cooldown after any loss
Revenge trades destroy accounts. After any realized loss above a minimum threshold, new entries are blocked until 08:00 EDT the following day. Risk-managing open positions stays unrestricted.
default: 1 loss → next session min-loss threshold configurable
04
No stop-loss widening
If a trader tries to move their stop further from entry — the single most destructive discretionary act in retail trading — RiskGuard cancels the widened stop and auto-restores the original in one round-trip. No gap in coverage.
on-chain restore trigger + limit orders TP/SL OCO aware
05
No pyramiding
Any order that adds to an existing position — or exceeds it and would flip direction — is cancelled at the pre-trade layer. Traders size once and manage what they have. Less tilt, longer survival.
flip-detection included reduce-only bypass
§ 04 / integration

Ship it in an afternoon.
Not a quarter.

One init call. Your order pipeline now has guardrails.

RiskGuard runs as a sidecar service against your DEX's public API. You forward new orders through our pre-trade endpoint; we validate against the trader's rule set and either pass-through or reject with a structured reason code. Everything else — state, fills, position tracking — is polled server-side.

No smart contracts to deploy. No custodial trust required. The trader retains key control; we only cancel their own orders using a delegated signing key with strict scope.

  • Hyperliquid adapter shipping now · dYdX + Drift + GMX Q2
  • REST + WebSocket · <3s cancel-to-acknowledgement
  • Webhooks to your backend on every rule fire
  • Per-trader rule overrides via JSON or your admin UI
  • Audit log exportable for MiCA / CFTC compliance
typescript
python
curl
// 1. Install
// npm install @riskguard/sdk

import { RiskGuard } from '@riskguard/sdk';

const guard = new RiskGuard({
  dex: 'hyperliquid',
  apiKey: process.env.RG_API_KEY,
  rules: {
    maxDailyLossPct:      5.0,
    maxDailyProfitPct:   100.0,
    cooldownAfterLoss:   true,
    noSLWidening:        true,
    noAddingToPosition:  true,
  },
});

// 2. Intercept orders before your matching engine
async function placeOrder(user, order) {
  const check = await guard.preTrade(user, order);

  if (!check.ok) {
    return { rejected: true, reason: check.rule };
  }

  return dex.submit(order);
}

// 3. Receive webhook on every rule fire
// POST /riskguard/webhook
//   { event: "daily_loss_lock", user, details }
§ 05 / pricing

Priced as infrastructure, not a surveillance tool. Your bigger moat is worth more than our margin.

Founding-pilot rate. First 3 DEXes only.

Two seats remain. Take one and the economics shift entirely: you pay at cost for six months while we re-run the retention model against your actual trader data, rate locked for life once the pilot converts, and if month-90 cohort fees don't move you walk and we delete the agent keys that afternoon. The next three customers pay Tier II rate from day one.

$0 setup · Tier-II SDK, all adapters
At-cost infra billing · months 1–6
Rate locked for life on conversion
90-day walk clause · full guarantee
Same-week pilot start · call Monday, live Thursday
Claim a founding seat →
Tier III
Long tail
< $50B annual volume · launch-stage DEXes
$750/mo
Flat · or 0.0005% rev-share
  • Hyperliquid adapter
  • All 5 core rules
  • Shared dashboard
  • Community Slack
  • Monthly rollup report
Start pilot →
Tier I
Hyperscale
> $500B annual · Hyperliquid, Aster, Lighter class
$10K+ /mo
Enterprise · co-branded · SOC 2
  • Everything in Tier II
  • Co-engineered rule extensions
  • White-labeled for your traders
  • On-call eng + weekly review
  • SOC 2 Type II + pen-test reports
  • MiCA / CFTC compliance package
Enterprise →
§ 06 / honest answers

The questions a skeptical DEX founder would actually ask.

Won't this just reduce our volume?
Short-term: yes. You'll see a few percent fewer tilt-trades. Mid-term: no, because the traders you keep are compounding. A trader who survives month one places 10× more lifetime orders than one who blows up in week two. Our pilot structure is explicitly designed around 90-day cohort volume, not daily snapshots. If the data doesn't move, you don't renew.
What if Hyperliquid just builds this themselves?
They might — and we'd cheer. But Hyperliquid isn't going to build configurable guardrails for every DEX downstream of them, and they're not going to white-label for your platform. We're not competing with Hyperliquid's liquidation engine; we sit one layer above, where your traders already want a seatbelt. If Hyperliquid ships an SDK that makes us obsolete in three years, we'll have already shipped to the other twenty.
Is this regulated?
Not today. But the CFTC's 2026 perpetuals framework will almost certainly require position limits and loss controls on US-facing venues, and MiCA already does in the EU. Platforms that have guardrails in place become the compliant choice by default. We're shipping the audit trail most regulators are going to demand within 18 months. Integrating now puts you ahead of the rulemaking, not chasing it.
How is this not just a stop-loss?
A stop-loss is a single trader-set order that can be cancelled, moved, or ignored. RiskGuard is a platform-enforced policy that runs server-side and cancels the trader's attempts to undo their own discipline. The SL-widening rule is the canonical example: a trader moves their stop further from entry (the classic tilt move) — we cancel the widened order and restore the original in one round-trip. A regular stop-loss has no immune system. We are the immune system.
What's the real integration time?
On our side: under 48 hours to running pilot. The RiskGuard daemon is already live on Hyperliquid mainnet. The SDK is a thin REST wrapper around the engine — your backend POSTs each new order to /v1/pretrade, we return { ok: true | false, rule, reason }, you pass through or reject. Median integration on a motivated engineer's side: one afternoon. Full production rollout with your admin dashboard and webhook wiring: three to five business days. Call us on a Monday, risk-managed cohort live by Thursday.
Who owns the trader's keys?
The trader. Always. RiskGuard operates via a scoped delegated signing key that can only cancel orders and place reduce-only stops on the trader's own account. It cannot open new positions, transfer funds, or access funding. Standard Hyperliquid agent-wallet pattern; security model is identical to the Hyperliquid UI itself.
What's the honest MRR of this business today?
Zero — by design. We're pre-revenue because we're closing with design partners before a public launch, not because no one's buying. The core engine is live on Hyperliquid mainnet right now, managing real trader accounts. We've had pilot conversations with 8 DEX founding teams. The first three founding seats get 6 months at cost with the rate locked for life — because we want three customers to build with, not three hundred to explain to. If that sounds like the wrong tradeoff for your DEX, we're probably not the right fit yet.
FOUNDING PILOT · 2 OF 3 SEATS OPEN

30 minutes.
Your actual numbers.
No pitch deck.

We run the retention model on your real trader cohort on the call and walk through the Hyperliquid instance live. If you're not convinced by minute 10, we say so and save us both the follow-up. Pilot terms sent same day if there's a fit.

90-day pilot guarantee. If your cohort's cumulative fees don't move relative to baseline, you cancel at 90 days and we delete the agent keys the same afternoon. You keep every integration artifact. Our downside — not yours.