The referral system - 0.2% volume rebate, paid in full
Atomic now pays referrers 15% of the trading fees their invitees generate, with no caps and no cliff. The dashboard surfaces what you earned, what your referrals paid, and a per-month breakdown across trading and lending. Here's how the math and the plumbing work.
The economics
Atomic charges traders 0.2% of notional volume on every position open and close. That fee splits three ways: liquidity providers, the protocol's revenue share, and - as of this release - referrers.
A referrer earns 15% of the trading fees their invitees pay. In other words: payout = volume × 0.002 × 0.15 = volume × 0.0003.
| Invitee's traded volume | Fees they paid (0.2%) | Your cut (15% of fees) |
| $10,000 | $20 | $3 |
| $100,000 | $200 | $30 |
| $1,000,000 | $2,000 | $300 |
| $10,000,000 | $20,000 | $3,000 |
No caps. No cliff. No vesting. Payout is in USDC.e, the same asset margins are denominated in.
We tested two splits in private rounds: 10% and 20%. At 10%, referrers stopped sending traffic after the second month. At 20%, the protocol's share was thin enough that it didn't make sense in pools where utilization sat below 60%. 15% landed where both sides sustain.
What's tracked
Per invited trader, the system keeps a running ledger of:
- Trading volume - sum of position open + close notional.
- Lending volume - USDC.e-equivalent value of any deposits the invitee made into the lending pool.
- Fees paid - exact USDC.e the invitee sent in trading fees.
- Your earned share - 15% of those fees.
Lending volume is converted to USDC.e-equivalent using the score system's minReward ratios, so a WBTC deposit isn't double-counted at face value. The conversion path is in helpers/refers.js on the server side.
The dashboard
Two surfaces, one for invitees and one for admins.
Referral dashboard (per user)
Lives at app.atomic.green/refer. Four stat cards across the top:
- Referrals Active - unique addresses that traded in the selected period.
- Trading Volume - sum across all your invitees.
- Fees from Referrals - what they paid in fees.
- You Earned - your 15% cut.
Below, a table lists each referral with their per-month breakdown of trading and lending volume. The period selector in the header switches the entire view between All time and any individual month back to launch.
Admin payouts (internal)
A separate /admin/referrals view aggregates the same data across all referrers. Three top-level cards: This month payout, All-time payouts, All-time fees collected. Below, collapsible blocks per month show every referrer with their address, count of active referrals, volume, fees, and computed payout. This is what we use to sign payouts at month-end.
The plumbing
Two endpoints back the dashboard:
POST /refer/getReferInfo
→ owner's referrals list, with per-referral monthlyBreakdown[]
→ used by app.atomic.green/refer
GET /admin/referral-payouts
→ all referrers, all months (last 12), with payout math
→ used by app.atomic.green/admin/referrals
→ wallet-gated to admin addressesThe monthly breakdown is computed by grouping positions.start_time and positions.end_time into UTC months, attributing each trader's volume to their referrer through the refers mapping, and tallying. Lending entries are grouped by lendings.time and normalized through the price helper.
Referral attribution is currently server-side - it lives in the API that backs the dashboard. The contracts don't enforce it. Payouts are settled manually at month-end from the recorded ledger. Moving attribution into the contract is on the roadmap, but the current model is faster to iterate on while we tune the split.
Getting started
- Connect your wallet at app.atomic.green/refer.
- The page generates a referral link tied to your address - share it.
- Anyone who connects through your link is bound to you the first time they trade.
- Earnings start accumulating from the first fee they pay. The dashboard updates in near-real-time (the polling cycle is 60 s).
- Payouts are settled monthly, last business day, in USDC.e to your referring address.
What's next
We'll publish a follow-up after the first full month with realized payouts, distribution shape (top-10 vs long-tail referrers), and any tuning we apply to the split. The schema for monthlyBreakdown will stabilize after that retro - if you're integrating against the API, watch the changelog for the version bump.