Stop babysitting your liquidity positions. Maltea's risk engine picks the optimal range on Orca or Raydium, then an autonomous vault keeps it centered, compounds the fees, and protects you from impermanent loss. Non-custodial, self-funded, always. Point a quantitative risk engine at any Orca or Raydium CLMM pools get a verdict, a best range and a strategy in seconds, then let a non-custodial vault keep the position on-plan: recentring, compounding and defending against impermanent loss while you keep custody of your funds.
Paste any pool — get fees, IL, LVR and a 0–100 verdict.
A non-custodial vault opens the position at the best range for you.
A keeper recentres hourly and compounds fees so IL stays bounded.
Live state, fees and price history from Meteora, Orca and Raydium — normalized into one model.
Volatility regime, fee density, LVR and a 4,000-path Monte Carlo produce a range, exposure and a 0–100 verdict.
The keeper recentres the range and hedges directional drift so impermanent loss stays bounded — target ≥80% mitigated.
Fees compound, the position tracks the model, and you can see exactly why at every step.
Maltea models impermanent loss at entry, then actively defends it — recentring the range and hedging directional drift to keep drawdown bounded. Target: ≥80% of IL mitigated versus a passive position.
Scans width after width and picks the one with the best net-vs-hold APR while staying in range — with capital efficiency and concentration for each.
GBM price paths give P(beat hold), P(profit) and a 5% Value-at-Risk — fees and IL simulated together, not bolted on.
Every verdict weighs fee income against loss-versus-rebalancing — the real cost of LPing.
Meteora, Orca and Raydium normalized into a single model — paste an address, done.
Deposit into a vault; the keeper recentres and hedges. Your keys, your funds, on-chain.
Execution roadmap
The current product helps users decide where and how to LP. The next layer is wallet connection, one-click CLMM position creation, position tracking, manual rebalance, and finally vault-based recentering with keeper automation.