Choose the execution mode that fits your testing goal. Start isolated, then opt into competition when you are ready.
Clean-room benchmarking
Real-world competitive testing
How agent orders interact — or don't — depending on the execution mode
The shared orderbook exposes behaviors invisible in isolated testing
Agents prone to bot races are exposed — those without proper race protection lose money fast in competitive matching.
Smart agents leverage weaknesses in others' strategies. The shared orderbook mirrors real exchange dynamics.
Agents that over-trade or fall for race conditions face realistic throttling and fill constraints.
In the real world, optimizers trade against each other — now you can test that before you deploy.
BESS + PV, BESS + Wind, BESS + Consumer — behind a single shared grid connection.
PEXim tracks each asset independently and enforces the grid export cap.
Charges at hours 8-13 when solar exceeds grid cap headroom
Discharges hours 15-20 when PV fades and prices rise
Combined output never exceeds 10 MW shared connection limit
Discharges during solar peak hours 8-13, stacking on top of PV output
Combined output exceeds 10 MW for 5 consecutive hours (09-13)
Each quarter-hour of violation is settled at imbalance price — typically 2-5x spot
Combined output exceeds shared connection limit → imbalance costs per quarter-hour
Actual PV/wind output differs from forecast → deviation settled at imbalance price
BESS dispatched below minimum SOC → infeasible volume charged at imbalance rate
You control who you compete against. Shared environments require explicit consent from all participants.
Every agent starts in its own isolated replica. Your strategy stays yours.
Choose to join a shared orderbook. Test against your own fleet first, or invite specific competitors.
Multiple owners can opt in together. Clear rules, fair conditions, transparent results.