Forward Testing EA: The Demo-to-Live Checklist
Forward testing EA is the disciplined bridge between a promising backtest and a live account. A backtest tells you how a strategy performed on historical data; a forward test tells you how it behaves in the real market, with real spreads, real slippage and real news. This guide covers why forward testing matters, how to set up a demo account and journal, what to record over a three-to-six-week observation period, and the checklist to work through before going live. It is part of the AlgoTM documentation hub and builds on the beginner playbook.
Forward testing EA: why it matters
Backtesting is indispensable, but it has well-documented blind spots. Historical tick data contains gaps, spread assumptions rarely match live conditions, and optimisation can accidentally over-fit a strategy to past price movements. Forward testing an EA on a demo account closes those gaps: the strategy trades in real market conditions, in real time, against the same broker infrastructure you will use with real money.
Forward testing also surfaces the problems no backtest can show. Platform disconnects, order-send failures, licence issues, weekend gaps and behaviour around high-impact news events all appear only in live conditions. A forward test is the cheapest way to discover these problems, and it has a less obvious benefit: it builds your understanding of the strategy. By the end of the observation period you should be able to explain what the EA does, when it loses money and why. If you cannot, you are not ready to trade it live.
Forward testing EA: setting up the demo account
Use a demo account from the same broker you intend to trade with live, funded to approximately the balance you plan to deposit, with the same leverage. Spreads, execution speeds and server locations vary between brokers, so testing on one broker and trading on another invalidates the exercise. Use the exact EA build and set file you will deploy live, and change nothing mid-test unless you are prepared to restart the observation period.
Create the trading journal before the first trade executes. A spreadsheet with one row per trade is sufficient: date and time, symbol, direction, entry and exit prices, lot size, swap and commission, net profit, drawdown at the time of the trade and a notes column. Use the position size calculator to fix the risk per trade in advance so the journal measures strategy performance rather than your discretion.
Forward testing EA: the observation period: Three to Six Weeks Minimum
Run the forward test for a minimum of three to six weeks. That is the shortest window that reliably covers a range of market regimes: quiet sessions and volatile ones, at least one major news cycle, weekend gaps and month-end flows. A three-day test on a scalper, or a single trending week on a mean-reversion EA, is statistically meaningless.
Three to six weeks is the floor, not the target. If your strategy trades infrequently, extend the period until you have a meaningful sample of trades; most strategies need at least thirty to fifty recorded trades before the equity curve tells you anything. For swing strategies trading a few times a month, that may mean two or three months of forward testing.
Forward testing EA: what to record
- Drawdown. Record the maximum peak-to-trough equity decline and compare it with the EA’s design parameters. If live drawdown exceeds the backtest by a wide margin, the strategy is behaving differently in real conditions.
- Win rate and profit factor. Track both across the whole period and in chunks. A strategy that wins steadily for three weeks and then gives it all back in two days has a behaviour you need to see.
- Behaviour in news. Note what happens around high-impact announcements: does the EA trade through them, widen its stops, or get whipped out? Decide in advance whether that behaviour is acceptable.
- Execution quality. Record slippage, requotes and failed orders. Consistent slippage beyond expectations changes the profitability picture and may point to a server-location issue.
- Operational stability. Note disconnects, platform crashes and whether the terminal was running when it should have been. If your computer cannot stay online, a VPS becomes a requirement rather than an option.
- Deviations from the backtest. The forward test does not have to match the backtest exactly, but unexplained differences, such as dramatically fewer trades or unusual trade frequency, need a documented explanation.
The Demo-to-Live Checklist
Work through every item below before funding a live account. Missing any single item is a reason to wait.
- The observation period has run for at least three to six weeks.
- You have recorded at least thirty to fifty trades, or the extended minimum appropriate to your strategy’s trade frequency.
- Maximum drawdown stayed within the strategy’s designed parameters.
- Behaviour around news events has been reviewed and consciously accepted.
- Execution quality matches the assumptions used in the backtest.
- Risk settings are confirmed with the position size calculator and match your account size.
- You have a plan for continuous operation, including a VPS if the strategy trades around the clock.
- You are funding the live account with capital you can afford to lose.
- You will start with a small lot size for the first ten to twenty trades.
- A monitoring schedule is in place before the first trade is placed.
The Decision Gate: What a Passing Test Looks Like
At the end of the observation period, run three simple tests. First, is the equity curve recognisably the strategy you backtested, with drawdown and trade frequency within tolerance? Second, did every loss come from a reason you can articulate — spread, news or trend change — rather than an operational failure? Third, is the journal complete enough that another person could follow your reasoning? If the answer to any is no, the EA does not pass the gate. Go back to configuration or continue observing; a live account will not fix a strategy that failed on demo.
If the EA passes, deploy with discipline: small size first, monitor the first ten to twenty trades against the forward-test baseline, and scale only after live behaviour matches demo behaviour. The full documentation hub covers deployment, monitoring and troubleshooting for every AlgoTM product.
Frequently Asked Questions
How long should I forward test an EA before going live?
At least three to six weeks, and ideally long enough to record thirty to fifty trades. The purpose is to see the strategy through varied market conditions, including news events and quiet periods, before risking capital.
What is the difference between backtesting and forward testing?
Backtesting runs the strategy against historical data, which is fast but limited by data quality and assumptions about spreads and slippage. Forward testing runs the same strategy on a demo account in real market conditions, revealing execution issues, news behaviour and operational problems that backtesting cannot show.
What should I record in my forward testing journal?
Every trade with date, symbol, entry, exit, lot size and net profit, plus running drawdown, win rate, profit factor, slippage and notes on behaviour around news events. The journal is the evidence you use to decide whether the EA is ready for live trading.
How do I know when an EA is ready for a live account?
Work through the demo-to-live checklist: a completed observation period, drawdown within design parameters, acceptable news behaviour, verified execution and a complete journal. If any item is missing, continue forward testing rather than deploying live.
Risk disclosure: Trading foreign exchange, commodities, CFDs and cryptocurrencies carries a high level of risk and may not be suitable for all investors. Past performance is not indicative of future results. AlgoTM provides trading tools and technology only and does not provide investment advice, portfolio management or guaranteed returns.