Strategy Tester: Run Forward Only for Positive Backtest Results?

[Enrique Enguix]

Hello everyone, good morning.

I have a question about the Strategy Tester.

When I optimize the parameters of an EA, I usually split the history into two parts: backtest and forward. The problem is that the tester runs the forward even for parameter sets that already gave negative results in the backtest.

What I’d like is that only the sets with positive backtest results are tested in the forward stage, so I don’t waste time with already discarded configurations.

Does anyone know if there’s a way to force the tester to behave like this?

Thank you!

[Eleni Anna Branou]

I don’t think that is possible, Enrique.

As far as I know, forward testing is carried out only on a portion of the backtest results, not on all of them.

For example, in my own optimizations, the Strategy Tester usually performs around 3,000 forward tests when the backtest results exceed 10,000.

[Enrique Enguix]

Eleni Anna Branou #:

I don’t think that is possible, Enrique.

As far as I know, forward testing is carried out only on a portion of the backtest results, not on all of them.

For example, in my own optimizations, the Strategy Tester usually performs around 3,000 forward tests when the backtest results exceed 10,000.

Eleni, thank you so much for responding as always.

It's strange, in my case, of the 10,000 results obtained in the backtest, 10,000 are tested again in the forward test.

Now I'm wondering if the backtest passes are being tested in the forward test (I'd like to assume they are).

Hahaha, so many years using the strategy tester, and now I have this question.

[Eleni Anna Branou]

Enrique Enguix #:

Eleni, thank you so much for responding as always.

It's strange, in my case, of the 10,000 results obtained in the backtest, 10,000 are tested again in the forward test.

Now I'm wondering if the backtest passes are being tested in the forward test (I'd like to assume they are).

Hahaha, so many years using the strategy tester, and now I have this question.

I assume that the Strategy Tester forward tests only the most promising backtest results, Enrique.

[Fernando Carreiro]

As per the documentation ... Strategy Optimization - Algorithmic Trading, Trading Robots - MetaTrader 5 Help

Forward testing is the repeated run of the best optimization results on a different time period.

The full optimization (slow or fast) of the Expert Advisor is conducted on the back testing period. After that 10% (in the full search) or 25% (in the genetic algorithm) of best runs are selected and then tested on the forward period.

There is a lower limit for the number of passes of forward testing. If the number of best runs is less than 256, the additional best runs are used for forward testing until their number reaches 256. If the number of all runs is less than 256, all of them participate in forward testing.

[Eleni Anna Branou]

Fernando Carreiro #:

As per the documentation ... Strategy Optimization - Algorithmic Trading, Trading Robots - MetaTrader 5 Help

Forward testing is the repeated run of the best optimization results on a different time period.

The full optimization (slow or fast) of the Expert Advisor is conducted on the back testing period. After that 10% (in the full search) or 25% (in the genetic algorithm) of best runs are selected and then tested on the forward period.

There is a lower limit for the number of passes of forward testing. If the number of best runs is less than 256, the additional best runs are used for forward testing until their number reaches 256. If the number of all runs is less than 256, all of them participate in forward testing.

That makes sense Fernando, thank you.

[Enrique Enguix]

Fernando Carreiro #:

As per the documentation ... Strategy Optimization - Algorithmic Trading, Trading Robots - MetaTrader 5 Help

Forward testing is the repeated run of the best optimization results on a different time period.

The full optimization (slow or fast) of the Expert Advisor is conducted on the back testing period. After that 10% (in the full search) or 25% (in the genetic algorithm) of best runs are selected and then tested on the forward period.

There is a lower limit for the number of passes of forward testing. If the number of best runs is less than 256, the additional best runs are used for forward testing until their number reaches 256. If the number of all runs is less than 256, all of them participate in forward testing.

Hello, good morning, Fernando.

Well, in my case, that's not the case. And I don't know why. There are the same number of passes in both tests. With genetic optimization, there are about 10,000 tests in both cases.

I use Custom Max as an optimization criterion, but I don't suppose that has any influence at all.

[Miguel Angel Vico Alba]

Enrique Enguix #:

Hello, good morning, Fernando.

Well, in my case, that's not the case. And I don't know why. There are the same number of passes in both tests. With genetic optimization, there are about 10,000 tests in both cases.

I use Custom Max as an optimization criterion, but I don't suppose that has any influence at all.

The optimization criterion has no effect on that. If you consistently see equal counts, it does not match the specification, so it is advisable to report it in the announcement thread of the build you are using.