Trading Styles & Strategy Development · Intermediate · 4 min
Forward testing and demo phase
You can plan a forward test on demo and with small live positions, compare its results with the backtest and decide when the rules are ready for normal size.
From past to present
A backtest answers one question: how would these rules have done in the past? Even an honest one cannot answer the next two: how will they do from here on, and how will you do when real money is on the line?
Forward testing bridges that gap. It costs time instead of money, and it catches many problems while they are still cheap.
Demo, small, normal
A forward test runs locked rules in real time, on prices that did not exist when the rules were written. That makes it the cleanest out-of-sample test there is: nothing in it can have been fitted.
A common path:
- Backtest on history, with out-of-sample data kept aside.
- Demo forward test. Trade the rules on a demo account with play money, exactly as written, for a set number of trades.
- Small live size. Trade real money at a fraction of normal size, for example a quarter of the planned risk.
- Normal size, only if the results of both stages are in line with the backtest.
Set the criteria before you start: how many trades each stage needs, which numbers you compare, and what result would send you back to the drawing board. MetaTrader 5's strategy tester has a related option: it can split the tested period and check optimized settings on a later, forward part.
What demo shows and hides
A demo shows whether the rules can be executed in real time: whether signals are clear, whether you see them in time, and how orders behave on the platform.
It hides two things. First, emotions: losing play money does not hurt. Lo, Repin and Steenbarger (2005) measured day traders' emotional reactions and found that traders with stronger emotional reactions to gains and losses tended to do worse. Second, some execution details: demo fills can be more generous than real ones, especially around news.
That is why the small live stage matters. If results drop sharply when money is real, the problem is usually execution, not the rules.
Common mistakes
- Skipping the demo because the backtest looked great. Overfit rules fail here cheaply.
- Changing rules during the forward test. Then it tests something else.
- Ending the test after a few good trades. Use the trade count you set in advance.
- Jumping from demo to full size. Real money changes behavior; let small size prove it first.
The takeaway
A forward test runs the locked rules in real time, first on demo, then with small money. Only results that survive both deserve normal size.
Check yourself
Quick quiz
4 questions. Instant feedback, nothing is saved.
-
1Tap the step that tests the rules in real time without risking money.
Tap the chart.
A demo forward test runs the locked rules on live prices with play money.
-
2What is the key difference between a backtest and a forward test?
Real-time data cannot have been used to fit the rules, so it is the cleanest out-of-sample test.
-
3Why move to small live size before normal size?
Many traders follow rules perfectly on demo and differently when money is at stake.
-
4The forward test is much worse than the backtest after 80 trades. What is a sensible reading?
A large gap often means the backtest was overfit or the rules are not being followed.
of 4 correct.