From Demo to Live: The EA Go-Live Checklist

Going live with an EA is not a single switch - it is a sequence: confirm the evidence, replicate the setup exactly, start at minimum size, and watch the first weeks like a hawk. This checklist walks through each stage so the gaps get found on demo terms, not at your account's expense.

Before anything: is the evidence actually there?

• Proper validation, not just a backtest. The system should have survived out-of-sample testing - a good-looking in-sample run alone is how traders fund broken systems, as covered in backtest vs forward test.
• A meaningful demo period. Long enough to confirm the EA trades as tested - frequency, behaviour, execution - while accepting that a short demo cannot statistically prove the edge. Know the real differences too: demo vs live trading.
• Known worst case. You can state the tested maximum drawdown and you genuinely accept a live result somewhat worse.

The go-live checklist

1. Replicate the environment exactly. Same platform version, same broker type if possible, same symbol (watch for broker suffixes like EURUSD.pro), same timeframe. Fresh install if needed: MT5 / MT4.
2. Carry the settings over precisely. Load the exact configuration that passed demo - a saved set file makes this one click instead of an act of memory.
3. Check every switch. Global Algo Trading / AutoTrading button on, per-EA permission ticked, happy icon on the chart. The full permission map: AutoTrading settings explained.
4. Start at minimum size. The first live weeks are about verifying execution - spreads, slippage, fill quality - not about earning. Trade the smallest size that produces real fills.
5. Confirm the first trades match expectations. Compare live entries against what the logic should have done. Check the Experts log for errors. Any mismatch: pause and diagnose before scaling.
6. Set up monitoring and uptime. A VPS or always-on machine, plus alerts, so the EA is not secretly offline for days - see monitoring EAs remotely.
7. Write the kill criteria now. The drawdown level and behaviour changes that mean "switch off", decided while calm. Details in reading strategy decay.
8. Scale in stages. Only increase size after a defined period of live behaviour matching the tested envelope - and keep total risk inside a sensible account-level budget.

Common mistakes

• Going live at full size on day one. The most common and most expensive shortcut.
• Changing settings at the moment of going live. New risk level means the demo evidence no longer describes what you are running.
• Judging the edge on week one. Early live weeks verify plumbing; the edge needs a sample.
• No exit plan. Deciding kill criteria mid-drawdown reliably produces the worst decision available.

The same bar for every system

This checklist applies to any EA, wherever it came from. Systems built in Nebula arrive with the validation stage already done - tested on unseen data before export - but they earn live deployment the same way: demo first, minimum size, monitored against expectations.

Trading carries a high level of risk. Nothing here is financial advice; past performance does not predict future results.