Knowledge baseFeatures

Expert mode, System Lab & cycles

When the guided scenarios aren’t enough — full manual control over every parameter and strategy, advanced lab tools, and how a grid runs cycle after cycle.

4 min read

Full manual control

The system builder normally guides you through a scenario. Expert mode (the full form) drops that scaffolding and exposes every parameter and every strategy at once — for when you know exactly what you want to build and the guided path is in your way. It is the same engine underneath; you simply see all the knobs.

System Lab

System Lab (available on the higher plans) adds advanced action tools on top — most notably auto-optimization, a grid-search that runs many backtests (up to a couple of hundred per job) to explore a parameter space for you. It is a research accelerator, not a promise: it finds what performed on history, which you then validate honestly.

Cycles & auto-restart

Grid-family strategies (Grid, DCA, Combo) work in cycles: the system averages in, reaches its take-profit target, books the profit, and closes the cycle. Two controls govern repetition:

  • Auto-restart — after a cycle closes, immediately begin a new one, so the system keeps working a range hands-free.
  • Trades limit — cap how many cycles it runs before stopping, so it doesn’t run forever unattended.

These apply only to the grid family. A one-shot strategy (Sniper, TWAP) has no cycle to repeat — the controls simply don’t apply to it.

The pitfall

Expert mode removes the guardrails of the guided flow — it will happily let you build something the scenario would have steered you away from. And auto-restart with no trades limit runs indefinitely: fine in a range, punishing if the market has broken out of it. Use expert mode when you genuinely know the trade-offs, and pair auto-restart with a trades limit and protection.

Next

Prove any expert build in Test Lab & Proof, and shape the grid itself in Grid shape.

Reading is free. So is testing.

Paper trading costs nothing and needs no card. Build a system, backtest it on real candles, and decide from the result.