The Pre-Automation Checklist: 9 Things to Verify Before Any Bot, EA, or Copier Touches a Live Account

Automation doesn't remove risk from trading. It removes your ability to react to risk in real time. That trade is worth making — a well-configured EA or copier reacts faster and more consistently than you do — but only if the boundaries were set correctly before it went live.

Every item below exists because a trader found out the hard way what happens when it wasn't. They group into three failure classes: the tool doesn't survive real-world conditions, the tool has no floor under its risk, or the tool keeps running when a human would have stopped.

Class 1 — Survivability: does it hold up outside a perfect connection?

1. You've run it on demo long enough to see a losing streak. A week of demo testing during a trending market proves the tool can open trades. It doesn't prove anything about how it behaves during a losing streak — does it size down, does it respect a daily-loss cutoff, does it keep trading into a drawdown the way you would if you weren't watching? If you haven't seen it lose five trades in a row on demo, you haven't tested it.

2. You know exactly what happens on a disconnect. Your broker terminal loses connection. Your VPS reboots for a Windows update. Your internet drops for four minutes. A disconnect is not a rare event over a long enough timeline — it's a certainty. What does your automation do in that gap: leave orphaned positions with no stop loss, resume cleanly, or double up on a trade it thinks never opened?

3. You've checked the broker-side math, not just the strategy math. Spread, commission, swap, and slippage compound differently across brokers and account types. A strategy that's profitable on a raw-spread ECN account can be a net loser on a standard account with wider spreads working against your typical holding time — pull your actual broker's cost schedule before assuming a backtest applies to your account.

Class 2 — Risk floor: can a bad trade become a bad week?

4. Every position has a stop loss, structurally, not by convention. "The EA always sets a stop" is a habit, not a guarantee. A structural stop means the position cannot exist without one — enforced by the tool, not assumed by the strategy. If a bug or a slipped fill can leave you holding size with no stop, that's the failure mode that turns a bad week into an account-ending one.

5. Lot sizing is tied to account equity, not a fixed number. Here's the arithmetic that matters: say you're risking 1% per trade with a 40-pip stop on a €10,000 account. At €10,000 balance, 1% risk is €100, which on a 40-pip stop works out to roughly 0.25 lots on a standard EURUSD contract. Grow that account to €15,000 and the same 1% risk is now €150 — but a fixed 0.25-lot size never adjusts, so you're now risking 0.67% instead of 1%. Shrink the account to €7,000 and that same fixed lot size is now risking 1.4%. A static lot size is a decision made once and wrong from the moment the balance moves; risk-percentage sizing is the only version that stays correct.

6. You've read the actual logs, not just watched the equity curve. An equity curve going up tells you the outcome. Logs tell you whether it got there the way you think it did — skipped signals, retried orders, silent errors that happened to net out fine this week but won't next week.

Class 3 — Enforcement: does the system stop itself, or does it wait for you?

7. Daily and weekly loss limits are enforced by the system, not by you checking a dashboard. If the plan is "I'll turn it off if it has a bad day," that plan depends on you being awake, present, and disciplined at the exact moment it matters — precisely the condition automation was supposed to remove. A real loss limit halts trading automatically and requires manual re-enable.

8. You have a kill switch that actually works under load. Test it while the tool is mid-cycle, not just when it's idle. A kill switch that only works cleanly when nothing is happening isn't a kill switch, it's a false sense of one.

9. You understand what the tool does with pending orders it didn't place. If your automation manages an account that also has manually-placed pending orders, does it touch them, cancel them, or ignore them? Mixed manual-and-automated management on the same account is one of the most common sources of "I don't know why that trade closed" support tickets.

Key takeaways

• Group your own pre-launch checks into the same three classes: does it survive a disconnect, does every trade have a real risk floor, does it enforce limits without you.
• The riskiest gap is usually Class 2 — a tool can look perfectly reliable in Class 1 and 3 and still have no structural stop loss underneath it.
• None of this is specific to one product. It applies whether you're running a copier, a signal-execution bot, or a fully mechanical EA.

WARNING

None of this replaces your own judgment about position sizing and overall portfolio risk. Automation executes a plan faster and more consistently than a human — it does not know whether the plan itself is sound. That part is still yours.

FAQ

Do I need to re-run this checklist for every EA or copier I add, or just once per account? Per tool. Two EAs on the same account can each pass Class 2 individually and still combine into more total risk than you intended — check aggregate exposure whenever you add a second automated system to an account.

What's the minimum demo period before I trust the results? Long enough to include one losing streak and one high-volatility news event, not a fixed number of days. A quiet week proves nothing; a week that included both is worth far more than a month of calm price action.

Does this apply to the copier hub or just EAs? Every item on this list — structural stops, equity-based sizing, an enforced kill switch — applies to account-to-account copying exactly the same way it applies to a strategy EA. The mechanism executing the trade doesn't change what's required before you trust it with real money.