Lighter LLP — what a perp DEX market-making vault really earns
Data as of 3 Oct 2026.
Verdict: the yield is real, but it isn't stablecoin yield. LLP's share price confirms 17.2% annualised over 90 days against 3.6% on Aave USDC (Ethereum). Over a full year it's 7.8%, because the window includes the 10 Oct 2025 crash: −5.35% in a day and 117 days back to the high. The real story isn't the vault but the entry ticket: deposits are only possible against staked LIT at 1 LIT per 10 USDC, so at $3.50 that's 35% of the deposit held in LIT. The only fair comparison with a stablecoin deposit is that package.
What it is
LLP (Lighter Liquidity Provider) is the vault of the Lighter perp DEX, which "provides liquidity to Lighter, runs market making strategies, and handles liquidations" (API description field). It's the exchange's market maker and liquidator: it earns spread and liquidation flow and loses when traders win or the market gaps. No operator fee (operator_fee 0). Deposits are in USDC; the share price moves with the vault's P&L.
LLP realised yield by share price (Lighter API, pool_info.share_prices, 625 daily points since 16 Jan 2025, compounded annualisation) vs Aave USDC on Ethereum (on-chain supply index, 22 Sep 2026).
Findings
1. Realised yield by share price 🔴
| Window | Share price change | Annualised (LLP) | Aave USDC Ethereum | Spread |
|---|---|---|---|---|
| 30 days | +1.25% | 16.3% | 3.76% | +12.5pp |
| 90 days | +3.98% | 17.2% | 3.62% | +13.6pp |
| 180 days | +6.54% | 13.7% | 3.77% | +9.9pp |
| 365 days | +7.78% | 7.8% | 3.58% | +4.2pp |
| since 31 Dec 2025 (post-token) | +8.53% | 11.5% | — | — |
| since launch (16 Jan 2025) | +261% | 112% | — | not comparable: the vault was tiny in its first months |
The 2026 monthly series is steady and moderate: −0.02% (March) to +1.85% (August), mostly 0.8–1.0% a month — ~10–12% annualised in a calm market. The API's live APY of 14.83% is closer to reality than the ~34% quoted by secondary sources, but it's still a snapshot.
The "~34% vs 17%" gap is two dates, not two measurements: by share price, 90-day annualised yield on 13 Nov 2025 was 34.7%, matching onchaintimes (13 Nov 2025). The vault's yield has halved in a year — a finding, not noise.
2. The entry ticket — a third of the deposit in a volatile token 🔴
Per the docs, each staked LIT allows a deposit of up to 10 USDC into LLP; unstaking LIT has a 3-day cooldown; staking pays a fixed 6% a year (docs.lighter.xyz/about-lighter/lit-utility; introduced 14 Jan 2026, The Block). LIT is $3.50 on two sources (CoinGecko and Lighter's LIT/USDC spot, 3 Oct). So every $100 in LLP needs $35 in LIT.
LIT daily closes (CoinGecko): 31 Dec 2025 $2.70 → low on 31 Mar 2026 of $0.81 (−70%) → high on 24 Sep 2026 of $5.32 → $3.50 on 3 Oct 2026 (−34% in 9 days). The ratio is set in tokens, not dollars, so a falling LIT doesn't push a depositor out of LLP — but the LIT itself loses value. Careful: LITUSDT on Binance is a different token (Litentry).
3. Worst episodes by share price 🔴
- 10–11 Oct 2025: −5.35% in a day; 117 days to recover the high (5 Feb 2026). Context: a ~$19bn market-wide liquidation cascade and a 4.5-hour Lighter outage caused by database growth. ADL didn't fire — the vault ate the hit. LLP depositors were compensated in points only (~$21.5m of vault losses).
- March 2025: max drawdown −8.9% (2→5 Mar), worst day −6.73%; the vault was several times smaller then, so it's weak precedent.
- Since 5 Feb 2026: worst day −0.165%, daily standard deviation 0.085%, 51 down days out of 180. No major event since the vault was split into isolated strategies — the new design's tail is untested.
- The series is one daily snapshot at 20–21:00 UTC; intraday drawdowns are invisible.
4. The rail: a Stage 0 ZK rollup with instant upgrades 🔴
L2BEAT (3 Oct 2026): SNARK proofs, data posted to Ethereum as blobs, an escape hatch ("desert mode": if the operator is down for 14 days, exit via a ZK proof of balance). But the contracts are "instantly upgradable", so there's no exit window against an unwanted upgrade. The standard upgrade delay is 21 days (3/5 multisig), but a separate 4/7 multisig can cut it to zero. TVS is $1.19bn, 44.6% of it LIT itself. The escape hatch doesn't protect against a bad upgrade: the rules can change before a depositor gets out.
5. Who runs the vault and what's in it
"Community-owned and protocol-run": the protocol runs the strategies; the operator's share is ≈0.004% of units — effectively no operator skin in the game. Lighter's treasury put in $7.5m in December 2025 (≈9% of today's vault, if still there — unverified). After 10 Oct the vault was split into isolated strategies ("risk and losses are isolated at the strategy level"), but the docs don't say which strategy covers which markets. The API lists 8 strategies: the largest at $54.1m (64% of the vault), then $13.6m and $10.2m.
Vault positions on 3 Oct: 198 markets, ≈$20m gross exposure on $85.1m of collateral (≈0.24x), net −$3.4m (short). Largest: QQQ +$4.8m against US100 −$4.7m (nearly hedged), ETH −$1.1m, XAU −$1.05m. The vault also carries FX, indices and gold, not just crypto.
6. The vault has shrunk roughly sixfold
About $500m in November 2025 (onchaintimes, single source) vs $85.1m now (API total_asset_value). The likeliest cause is the LIT staking requirement from 28 Jan 2026 (an inference, not measured). A smaller vault means more profit per dollar from the same spreads — and a thinner buffer for one cascade.
7. Withdrawals
Public pools have no lockup and funds can be withdrawn at any time (docs, public-pools); the API's pending_unlocks is empty. LIT unstakes after a 3-day cooldown. Incident: on 30 Dec 2025 the prover fell ~4 hours behind the sequencer and withdrawals failed with "Too many L2 Withdrawals"; they resumed on 31 Dec and $250m, ~20% of TVL, left in a day (CoinDesk).
8. Jurisdiction and audits
Lighter's terms bar persons located in or resident in the US, Canada, the UK, China, North Korea, Russia, Ukraine, Cuba, Iran, Venezuela, Syria (the preamble adds Sudan, Belarus and Myanmar) and sanctioned countries. The test is location and residency; citizenship isn't mentioned. There's no specific clause on freezing withdrawals, only a general right to suspend interface access.
Audits: docs/security-audits hosts 9 unlabelled files; secondary sources say ZK Security and Nethermind reviewed the circuits. The reports weren't opened; dates and scope are unconfirmed.
What it actually pays
- The vault alone, ex-LIT: +9.9–13.6pp over Aave on 90–180 days and +4.2pp over a year with one major event. A 10 Oct-scale event takes ~5.4% in a day — about 4–5 months of the current premium (≈0.9–1.2pp a month). A fair expectation for a year with one such event is 3–5pp over Aave; without one, 8–13pp.
- The package as it really is (vault + 35% in LIT): the LIT leg earns a fixed 6% staking reward, but LIT traded in a −70%…+560% range off its low in 2026. Per $100 in the vault, a −30% LIT move (like the last 9 days) costs $10.50 — almost a full year of vault income (~$12). Over a horizon of months, LIT price action dominates vault yield. As a stablecoin-deposit substitute, the package fails: it embeds a directional bet on LIT.
- Can the LIT leg be hedged? Lighter lists a LIT perp ($23.2m daily volume, OI ≈$9.8m). Shorting the perp against the staked amount removes LIT price risk in theory; funding on 3 Oct was positive (shorts get paid), but the funding period in the API field isn't documented. The cost: another leg on the same venue and liquidation risk on the short if LIT spikes. A hypothesis, not a ready strategy.
- As a "vault vs traders" product, LLP is a cousin of HLP and JLP. Published Sharpe ratios for such vaults are partly survivorship; halving the API's 3.85 gives ~1.9.
What would prove this wrong
- A cascade the vault can't de-risk in time: a daily share-price drop worse than −5.35% or a drawdown deeper than −8.9%.
- A zero-delay contract upgrade (4/7 multisig): the escape hatch doesn't help.
- LIT below $0.81 (the 2026 low): the vault + LIT package is negative over a year even with 15% from the vault.
- Forecast for 31 Dec 2026: LLP share price between 0.003675 and 0.003790 vs 0.0036103 on 2 Oct 2026 (+1.8–5.0% for the quarter, ~7–21% annualised). The low end breaks on a 10 Oct-scale event, the high end if the 2025 regime returns.
- Forecast for 31 Dec 2026: LLP collateral stays below $150m (now $85.1m) — the LIT staking requirement caps inflows.
What we don't know
- Depositor concentration: the API doesn't list unit holders, and the treasury's current share isn't verified.
- How markets are split across the 8 strategies — so how much of the vault carries crypto tail risk versus FX and indices.
- How isolated strategies behave in a 10 Oct-scale cascade — none has happened since February 2026.
- Whether the 1:10 LIT staking ratio will hold.
- Cumulative return since launch as of 13 Nov 2025: +228% per the API vs "122% since inception" per onchaintimes; the cause isn't established. The key series has a single primary source — the Lighter API — since there's no on-chain share contract.
Research for information only. Not investment advice.