Skip to main content

Validation And Forward Testing

Validation is the step between a historical test and real account use. Its purpose is to check whether a saved setup behaves sensibly outside the data used to create it.

Forward testing does not prove that a setup will be profitable. It helps you find obvious problems before using the EA on a live account.

warning

Historical tests, out-of-sample tests, demo tests, and small live tests are all limited. None of them guarantees future results.

Why Validation Matters​

Backtests are controlled simulations. Live trading includes conditions that a backtest may not fully capture:

  • changing spreads
  • slippage
  • broker execution delays
  • rejected or modified orders
  • symbol specification differences
  • VPS or internet interruptions
  • market conditions that differ from the test period

Validation is used to compare the EA's expected behavior with what happens under less controlled conditions.

Suggested Validation Workflow​

  1. Optimize or choose settings on one period of historical data.
  2. Save the settings as a .set file.
  3. Test the same settings on data that was not used during optimization.
  4. Run the same setup on a demo account.
  5. Review execution, trade frequency, drawdown behavior, and order handling.
  6. Only consider live use if you understand the risks and the setup behaves as expected.

Do not keep changing settings during validation. If you adjust the inputs after seeing validation results, start a new validation cycle.

Out-Of-Sample Testing​

Out-of-sample testing means testing settings on historical data that was not used during optimization.

Basic Process​

  1. Choose an optimization period.
  2. Reserve a later period for validation.
  3. Optimize only on the optimization period.
  4. Lock the selected settings.
  5. Run a separate backtest on the reserved period.
  6. Compare the result behavior, not just the headline profit number.

What To Review​

Check whether:

  • the EA still opens and closes trades normally
  • trade frequency is in a similar range
  • drawdown behavior is understandable
  • losses are not caused by one narrow market event only
  • results are not dependent on one unusually large trade
  • spread and commission assumptions are realistic for your broker

Warning Signs​

Review the setup carefully if:

  • the test depends on very few trades
  • one or two trades explain most of the result
  • drawdown is much larger than expected
  • the setup only works on one narrow date range
  • small input changes produce completely different behavior
  • broker costs change the result materially

Demo Forward Testing​

Demo forward testing runs the EA on live market data without using real funds.

Demo Setup Checklist​

Use the same conditions you plan to use later:

  • same broker if possible
  • same account type where possible
  • same symbol name
  • same timeframe
  • same .set file
  • same VPS or computer setup
  • same trading hours and filters

What To Monitor​

During the demo test, monitor:

  • whether trades open when expected
  • whether stop loss and take profit values are placed correctly
  • whether spread filters block trades at sensible times
  • whether order errors appear in the Experts or Journal tab
  • whether alerts and notifications work
  • whether VPS or platform interruptions occur
  • whether the EA behaves consistently after MT4 restarts

How Long To Test​

There is no single correct duration. A setup that trades often can be reviewed sooner than a setup that trades rarely.

As a practical rule, do not judge a setup from only a few trades. Wait until you have enough activity to understand its normal behavior, losing streaks, and broker execution conditions.

Small Live Testing​

Small live testing is optional and should only be considered after the user understands the EA, the setup, and the risk.

If you choose to do this:

  • use small position sizing
  • use money you can afford to risk
  • keep the same settings used during validation
  • monitor the account closely
  • stop the test if the EA behaves differently from the demo test
  • do not increase risk because of a short good period
danger

Live testing uses real funds. BananaEA is not financial advice, and you are responsible for your own trading decisions and risk management.

Validation Notes Template​

Use a simple record for each test:

Setup name:
EA version:
Broker:
Account type:
Symbol:
Timeframe:
Set file:
Test period:
Spread/commission assumptions:
Number of trades:
Main observations:
Execution issues:
Order errors:
Drawdown notes:
Decision:
Next review date:

The decision field can be simple:

  • continue testing
  • adjust settings and restart validation
  • reject this setup
  • pause because broker or platform conditions are unclear

Common Mistakes​

Avoid these validation mistakes:

  • optimizing on all available data and leaving no fresh period for validation
  • changing settings after every losing trade
  • judging a setup from one day or one week only
  • comparing demo and live results without checking spread and commission differences
  • ignoring order errors in the MT4 Journal or Experts tab
  • increasing risk before the setup has been reviewed properly