Demo to live: when is a gold EA ready for real money?
The most expensive moment in automated trading isn't a losing trade — it's going live too early. Here's the honest checklist for deciding when a gold EA has actually earned real money, and how to make the switch without betting the account on day one.
The three-stage path: demo → small live → scale
Skipping stages is how most EA stories end badly. Each stage answers a different question:
| Stage | Question it answers | How long |
|---|---|---|
| 1. Demo | Is the setup correct and does behaviour match the backtest? | 4–8 weeks, 20–30+ trades |
| 2. Small live | Do real spreads, slippage and fills change the results? | 4+ weeks at 0.01 lots / cent account |
| 3. Scale | Can I hold my risk rules as size grows? | Gradual — never a jump |
Notice what's missing: "the EA had a great week, let's go full size." A week of results — good or bad — is noise. The stages exist precisely because excitement is not evidence.
Stage 1: what demo testing must cover
Demo isn't about making imaginary profit — it's a systems check. Before the demo period counts, it should include:
- At least one NFP week and one FOMC week. Gold's character changes completely around big US releases; you want to see the EA (and its news filter, if it has one) handle them. Why this matters: trading gold during news.
- 20–30+ completed trades. Below that, results are statistically meaningless. If the EA trades twice a week, extend the demo period — the calendar serves the sample size, not the other way round.
- The exact live setup. Same broker, same account type, same settings, running on the VPS it will use live — not your laptop.
- Every stop-loss respected. If a single trade ran past where the strategy's stop should have been, stop and find out why before anything else.
The matching test: demo vs backtest
Here's the part most people skip. The point of demo isn't "did it make money" — it's "does live behaviour match the backtest?" Compare four things against your backtest results:
| Metric | Healthy sign | Warning sign |
|---|---|---|
| Trade frequency | Similar trades/week to backtest | Far fewer or far more trades than expected |
| Average win & loss size | Close to backtest averages | Losses noticeably larger than tested |
| Drawdown pattern | Dips within backtest's normal range | Deeper valleys than the backtest ever showed |
| Trade logic | Entries/exits you can explain from the strategy rules | Trades you can't explain |
Stage 2: why small live comes before real size
Demo servers fill orders perfectly. Live gold trading adds the costs that decide marginal strategies: real spreads (which widen at rollover and news), slippage on entries and stops, commissions, and occasional latency. None of these appear on demo — which is why the first live month runs at minimum size:
- Fund the account you actually plan to use — sized so a normal losing streak is survivable (our guide: how much money a gold EA needs).
- Run 0.01 lots (or a cent account) for 4+ weeks, whatever your eventual size.
- Compare the live month with the same weeks on demo: are spreads and slippage eating meaningfully into results?
- If live roughly matches demo — scale in steps, holding risk per trade at 1–2%. If it doesn't — the gap is the answer; investigate before adding money.
When to go back to demo
The demo→live door swings both ways. Drop back to demo or minimum size whenever something material changes:
- You change EA settings — new risk level, new sessions, new filters.
- You switch broker, account type (standard→raw, hedging→netting), or VPS.
- The EA does something you can't explain from its rules.
- Drawdown exceeds anything the backtest showed — pause, understand, then decide.
Re-testing after a change isn't paranoia; it's the same standard that made the first launch safe.
What this looks like with ForexGoldEA
ForexGoldEA is free via our partner brokers, so the demo stage costs nothing — you can run the full checklist before a single dollar is at risk. The EA's rule-based strategy, fixed stop-loss on every trade and no-martingale design make the matching test straightforward: trades are explainable, losses are capped, and drawdown stays visible. Set it up with the 5-minute install guide, demo it properly, then follow the small-live-first path above.
Frequently asked questions
How long should I demo test a gold EA?
4–8 weeks minimum, covering at least one NFP and one FOMC week, with 20–30+ completed trades. Extend if the EA trades infrequently.
What should demo results match before going live?
The backtest — similar trade frequency, win/loss sizes and drawdown pattern, with stops executing where the strategy says. Match, not beat.
Why do EAs perform worse live than on demo?
Real spreads, slippage, commissions and latency don't exist on demo. On gold these costs are meaningful, which is why small live size comes before real size.
How much should I go live with first?
Minimum size — 0.01 lots or a cent account — for 4+ weeks, regardless of your eventual capital. The first live month is a test, not a payday.
Can I skip demo and start with a small live account?
A cent account includes real execution costs, but demo first still catches setup mistakes at zero cost. Safest path: demo → small live → scale.
When should I go back to demo?
After any material change — settings, broker, account type, VPS — or whenever the EA behaves in a way you can't explain.
Bottom line
"When is my EA ready for real money?" has a boring, reliable answer: when demo behaviour has matched the backtest across 4–8 weeks including news weeks, and a month of minimum-size live trading has confirmed real execution doesn't break the edge. Every stage an EA skips is a risk transferred straight to your account balance. Let the robot earn its size — the gold market will still be there when it has.
Affiliate disclosure: we earn a commission when you open and fund an account through our partner links, at no extra cost to you.