How Long to Demo Trade on MT4: Losing Rules Looked Profitable After 10 Trades Up to 1 Time in 3, and a +1 Pip Edge Needed About 200
Key takeaways
- I took the real trades of six rules on 16,926 EURUSD 5-minute bars, with a spread and slippage, and resampled them 4,000 times to see what a demo of 10, 20, 50, 100 or 200 trades would have shown.
- All six rules lost 1.1 to 2.0 pips a trade over the full sample. After 10 demo trades they still looked profitable 25% to 33% of the time, and the mildest loser still showed a profit after 50 trades 16% of the time.
- A genuinely good system with +1 pip a trade looked like a loser after 20 trades one time in three, and after 100 trades one time in six. It took about 200 trades to show a profit 93% of the time.
- These M5 rules traded 22 to 45 times a week, so 200 trades is 4.5 to 9 weeks of demo. A two-week demo is too short to tell a small edge from none.
- Below: how to set up the MT4 demo so the numbers mean something, and a trade-count table to decide when to stop.
“Demo trade for a month first” is the usual advice. A month is a unit of time, and time is not what makes a demo result trustworthy. The number of trades is. I measured how many trades a demo needs, using the actual trade results from my backtest vs demo account test.
How did I measure demo length?
- Trades: six rules on EURUSD M5, June 3 to August 26, 2026: EMA 9/21, RSI 14, Bollinger 20/2, MACD 12/26/9, Supertrend 10/3 and Stochastic 14/3/3. Stop and target 15 pips, 261 to 536 trades each.
- Demo conditions: entry at the next bar’s open, 1 pip spread (2 pips from 21:00 UTC), 0.3 pips slippage on stops.
- A “demo”: N trades drawn at random from a rule’s real trades, 4,000 times for each N. I counted how often the demo showed a profit.
- A good system: none of the six had an edge, so I also made one: +1 pip a trade on average with the same spread of outcomes as the real rules (8.68 pips standard deviation per trade).
How often does a losing strategy look profitable on demo?
| Rule, full-sample result | 10 trades | 20 trades | 30 trades | 50 trades | 100 trades | 200 trades |
|---|---|---|---|---|---|---|
| Bollinger 20/2, -1.13 pips | 33.2% | 27.2% | 23.2% | 16.2% | 8.9% | 2.8% |
| MACD 12/26/9, -1.46 | 27.6% | 21.2% | 17.1% | 10.6% | 3.4% | 0.7% |
| RSI 14 30/70, -1.53 | 31.2% | 24.6% | 19.4% | 12.9% | 5.2% | 1.1% |
| Stochastic 14/3/3, -1.56 | 28.6% | 20.6% | 16.4% | 10.2% | 3.4% | 0.7% |
| EMA 9/21, -1.59 | 28.4% | 19.2% | 15.8% | 9.1% | 3.2% | 0.5% |
| Supertrend 10/3, -1.98 | 24.8% | 14.9% | 11.1% | 5.4% | 1.4% | 0.1% |
Share of 4,000 simulated demos that ended in profit. Every rule lost over its full sample.
So if you demo a typical losing indicator rule for 10 trades, you have roughly a one-in-three chance of seeing green and deciding it works. The closer a rule is to break-even, the longer it can hide: Bollinger, the mildest loser, still fooled one demo in six after 50 trades. The same thing showed up week by week: the fade rules had 3 or 4 green weeks out of 13 while losing overall.
How many trades does a real edge need to show?
| System with +1 pip a trade | 20 trades | 50 trades | 100 trades | 200 trades | 400 trades | 800 trades |
|---|---|---|---|---|---|---|
| Demo showed a profit | 67.3% | 77.1% | 83.6% | 93.0% | 98.3% | 99.8% |
| Demo showed a loss | 32.7% | 22.9% | 16.4% | 7.0% | 1.7% | 0.2% |
3,000 simulated demos per column, 8.68 pips standard deviation per trade.
This is the uncomfortable half. A good system, and +1 pip a trade after costs on M5 would be a very good one, looks like a loser after 20 trades a third of the time. Most people would quit it. The rule of thumb behind these numbers: trades needed is about (1.65 × spread of results ÷ edge)². With 8.68 pips of spread and a 1-pip edge that is about 200. Halve the edge and you need four times the trades.
How long is that in weeks?
| Rule, EURUSD M5 | Trades in 12 weeks | Per week | Weeks to 200 trades |
|---|---|---|---|
| Stochastic 14/3/3 | 536 | 45 | about 4.5 |
| MACD 12/26/9 | 515 | 43 | about 5 |
| Supertrend 10/3 | 397 | 33 | about 6 |
| EMA 9/21 | 380 | 32 | about 6.5 |
| Bollinger 20/2 | 363 | 30 | about 7 |
| RSI 14 30/70 | 261 | 22 | about 9 |
One position at a time, stop and target 15 pips, around the clock. Trading only a few hours a day cuts these counts and stretches the weeks.
A swing system on H1 or H4 trades far less. My RSI swing test got 32 H1 trades in 12 weeks from RSI 14, which means a meaningful demo of a slow system takes the better part of a year. For those, test on more history first and use the demo only to check execution.
How do I set up an MT4 demo so the result means something?
- Open the demo with the broker you will trade live. In MT4: File, Open an Account, pick the broker’s demo server, choose the same account type (standard or raw) as your future live account. Spread and commission differ by type, and on EURUSD M5 the spread is the biggest cost, as my tick data vs spread test shows.
- Use a realistic deposit. The amount you will actually fund, not the default 100,000. Size positions with the position sizing table so the percentages match live.
- Write the rule down before the first trade. Entry, stop, target, time exit, trading hours. Changing the rule halfway resets the count to zero.
- Take every signal. Skipping the ones that “look bad” turns a rule test into a test of your mood.
- Record in pips, not dollars. Account History tab, right-click, Save as Detailed Report. Add a column for pips per trade and one for the running average.
- Compare with the backtest, trade by trade count. After 50, 100 and 200 trades, put the demo’s average next to the backtest’s average with the spread included.
When should I stop the demo and decide?
| Trades so far | What the result can tell you |
|---|---|
| Under 30 | Nothing about profit. A losing rule showed green up to a third of the time. Only check that orders fill the way you expect. |
| 50 | A clear loss is worth listening to. A profit still fooled up to one demo in six on these losing rules. |
| 100 | A loss bigger than 1 pip a trade is very likely real. A profit is encouraging, not proof: a +1 pip system still showed red 16% of the time. |
| 200+ | Now the sign is usually right. A +1 pip system was green 93% of the time; the losing rules were green 3% or less. |
Two more checks before going live. First, a coin: random entries with the same stop and target made -1.14 pips a trade in my test, so a rule that cannot beat that is not a rule, as in rule vs random entry. Second, remember that a demo is the best case. Live fills on news and at the 21:00 UTC rollover are usually worse, per my spread measurement.
Frequently asked questions
How long should I demo trade before going live?
Count trades, not weeks. In this test a system with a real +1 pip edge needed about 200 trades to show a profit 93% of the time. For M5 rules trading 22 to 45 times a week that is 4.5 to 9 weeks.
Is one month of demo trading enough?
For a fast M5 system, a month gives roughly 95 to 190 trades, which is enough to reject a clear loser but not enough to prove a small edge. For an H1 swing system a month is far too short.
Can a losing strategy be profitable on demo?
Yes, often in the short run. Six rules that lost 1.1 to 2.0 pips a trade overall showed a profit on 25% to 33% of 10-trade demos.
How many trades do I need to trust a strategy?
Roughly (1.65 × the standard deviation of your trade results ÷ your average edge) squared. With 15-pip stops on EURUSD M5 and a 1-pip edge that is about 200 trades.
Should my demo account balance match my real account?
Yes. Use the deposit you will really fund and the same lot sizing, otherwise the risk per trade on demo tells you nothing about how the live account will feel.
Want a clean indicator to install right now?
It is my own enhanced DeMARK Trend Line indicator for MetaTrader 4 and 5. Non repaint, clean, and free.
Forex educator and indicator developer. I build and trade my own MetaTrader tools, and share the ones that genuinely help.