From Backtest to Live (Simulated) Trading

The last mile: take a strategy that earned its backtest and let it trade the platform's live simulated market. Upload the EA in the Trade Terminal, set its inputs, start it — from then on a Web Worker in your browser runs your OnTick against each new simulated tick, and every order it places goes through the same server route as your manual trades, opening and closing real (simulated) positions in your account.

The deployment checklist

Before starting any EA live, confirm all of it — this list is the course in miniature:

1. Backtested over 100+ trades spanning trends and ranges.
2. Robust parameters — plateau, not spike; sensible round values.
3. Out-of-sample pass on data that influenced nothing.
4. Stops attached — every OrderSend carries a stop loss. An EA without stops fails Course 03, no matter its backtest.
5. Sized by risk — lots derived from AccountEquity(), not a hardcoded constant.
6. Warm-up guarded — the Lesson 2 bug is checked at the top of OnTick.

Watch it like a scientist, not a fan

The first hours of live-simulated running are a test, not a verdict. Compare:

• Trades taken vs. the backtest's pattern — same frequency? Similar duration? An EA that suddenly trades twice as often is telling you something differs from the tested environment.
• Execution — intended entries/exits happening at intended prices?
• Behavior at extremes — the EA's worst minutes are more informative than its best hours.

Journal the EA's trades alongside your manual ones. A strategy that quietly degrades for weeks is the normal way edges end — variance or regime change — and the journal is where you notice.

Know the operating limits

• Your EA runs on your device, while the app is open. Close the tab and it pauses — nothing executes while the app is closed; reopen and it reattaches with your EA's saved variables. (Open positions keep marking to market server-side either way.) Pressing Stop is the explicit exit.
• The worker is single-threaded — keep OnTick lean. A slow or looping OnTick delays later ticks and any queued orders until it returns or you press Stop. Heavy computation belongs in preprocessing, not per-tick.
• Historical-data downloads (backtests, optimizations) are rate-limited on the platform — a nudge to act deliberately.
• One EA runs per account at a time: stop before starting another.

The honest expectations

A validated EA on this platform is a simulated result on historical data. The tester models the cost stack — spread, slippage, swap, $7/lot — the same cost stack the live terminal charges, but statically: real brokers widen spreads at news, gap through stops and shift regimes the backtest never saw, so its costs are a floor on real-broker realism. The professional stance: expect degradation, size for it (that 1% rule again), and let the live-simulated record — not the backtest — decide what (if anything) deserves eventual real capital.

You now have the complete loop: learn it, code it, test it, deploy it, measure it. Everything else is iterations and honesty.

By the numbers

The six checklist items — verified indicator warm-up, fixed risk input, a stop on every order, 100+ backtest trades, an out-of-sample pass, and a kill-switch — are worth exactly as much as your compliance with them. Two platform facts bound expectations: the backtester's static cost stack is a floor on real-broker realism, and the EA runtime is a Web Worker that pauses when the tab closes — nothing here trades while you're not watching.