Due diligence memo 02 · Prepared for Veris · 19 August 2026
JLP / USDC — Delta-Neutral Carry
Levered JLP on Kamino with the market exposure removed on Hyperliquid. Return driver: the pool's carry against its own traders.
Summary
JLP is the counterparty to every position on Jupiter Perps: its NAV carries the price move on its SOL/ETH/BTC/stable basket, a 75% share of platform fees, and the pool's P&L against traders. The strategy levers JLP on Kamino and shorts the basket on Hyperliquid to remove the first.
Risk dimension scores · 1 safest to 5 riskiest · overall = worst dimension
Smart contract
3
Three venues, deliberately split
Liquidity
3
Sized to 36% of available
Counterparty
4
Venue settlement precedent
Market
4
Carry is earned, not contracted
Oracle
3
NAV mark, verified to 0.008%
Upgradeability
4
Redemption switches are live
Leverage
3
1.75x with isolated hedge margin
Collateral
4
13.3% of NAV is a loan book
Operational
3
State fully readable on-chain
Monitoring
3
Custody-driven, not index weights
Return profile · $1,000,000 at 1.75x loop / 2.5x hedge
| Component | Rate | Basis |
|---|
| Delta-hedged carry, 365d | +9.32% | Out of sample, rolling betas, gross of costs |
| Delta-hedged carry, 2 years | +20.58% | Positive in eight of nine quarters |
| USDC borrow cost | −3.26% | 0.51 × capital at 6.41% |
| Funding received on shorts | +1.65% | 0.81 × capital, 365d mean |
| Net on committed capital | 8.53% | After entry, exit and rebalancing |
Conditions to fund · none met on 19 August 2026
| Condition | State on 19 August 2026 |
|---|
| Traders net losing, volatility elevated | Baseline regime — fails |
| Reserve utilisation below 90% | 92.5% — fails |
| USDC borrow under 5% | 6.41% — fails |
| Operational cycle proven live | $500k pilot not yet run |
Structure
VenuesKamino Multiply / Hyperliquid
Loop equity / hedge margin$677,507 / $322,493
JLP notional$1,185,637
USDC borrowed$508,130
Short notional$806,233
Volatile share hedged0.680 measured
Entry / liquidation LTV42.9% / 85%
Principal risks
| Risk | Dimension |
Score | Conf. | Assessment |
|---|
| The carry is earned from counterparties, not contracted | Market | 4 | M | No contractual floor. Measured at +43.6%, +21.3% and +8.5% annualised across 2024, 2025 and 2026 — positive throughout but decaying. |
| 13.3% of collateral NAV is an on-chain loan book carried at par | Collateral · Counterparty | 4 | M | The JLP USDC custody lends $103.5M against a $400M limit. If it impairs, NAV falls with no market move and the hedge produces no offset. |
| The hedge venue can settle open positions administratively | Counterparty | 4 | M | Hyperliquid validators voted in ~2 minutes in March 2026 to delist a market and settle all positions at a chosen price — for a hedge, a direct basis loss. |
| Redemption switches and freeze authority are live | Upgradeability | 4 | L | Per-custody flags can disable JLP redemption and are already in use elsewhere in the pool. Controllers and timelocks not yet resolved. |
| Two levered legs liquidate in opposite directions on different chains | Leverage | 3 | M | Loop liquidates on a −49.6% JLP fall, the hedge on a +37.0% rally. Hedge margin is pre-funded and isolated, so one failure cannot reach the whole account. |
| Entry draws on a single USDC reserve that runs near full utilisation | Liquidity | 3 | M | $508,130 of borrow against $1,408,134 available, or 36%. Exit is cleaner: repayment adds liquidity and redemption is 0.15% of a $775.8M pool. |
Prepared by Simone Taravelli for Veris. Confidential. Not investment advice. Scores run 1 to 5, where 1 is the safest and 5 the riskiest; 4 indicates significant unmitigated exposure warranting active monitoring; the overall rating is the worst dimension score, and 4 is the expected band for a levered cross-venue basis trade. Venue figures were read from Solana mainnet, the Kamino API and the Hyperliquid API on 19 August 2026. The hedged carry is computed from 905 days of JLP NAV and exchange closes, betas re-estimated out of sample on a trailing 180-day window; it is gross of financing and execution costs. The carry is paid by Jupiter Perps traders rather than contracted: it has no floor and is not statistically distinguishable from zero on the available sample. Past measurement is not a forecast.