Hi

Lets say you have two cases:
Case 1: You train algo A and B in a script as workshop6 and get the OptimalF parameteres of the assets for each algo.
Case 2: You train algo A and B in different scripts and then combine them by using the combine script.
Are the optimalF parameters similar or the same in both cases? Is there a rule in here like: Better train them together or separate?

The optimalF factors are the same. They depend only on the balance curves of the components.

I was thinking that algos interact between each other in the same script. Like one algo can close a trade opened by another algo. In that case they behave diferent together. Is it true?

You can control this behavior using the Hedge flag. The default is that opening a trade in an opposite direction, regardless of the algo, causes the open trade to close. You can control this so that only the same algo will close trades in the opposite direction, or you can allow full hedging.

Thanks.
I was aware of hedging but I did not noticed it can also be used to split the close of trades into algos. Definetly zorro team has taken everything into account.

The default is partial hedging. In the case of no hedging, the algos will affect each other, so you must then indeed either train the whole portfolio together or separate per asset, but not per algo.

Not sure now if it is better to train all algos together with just one asset, all assets with one algo or all together.
For example, if I expected to use 5 algos and 10 assets and Hedge=5 then the backtest should be with those settings or it can work just by doing each separate with Hedge = 1?

Training all algos together with one asset is the best way.

Hedge = 1 as well as Hedge = 5 causes the algos to affect each other, although with Hedge = 5 only the trading costs will be different.

Ok thanks.

Just to be sure. The workshop 6 should be actually trained just with one asset right?

I have another question tho. I see that there a couple of ways to use an indicator like an oscillator. One can enter trades in the peaks and valleys or after the indicator cross a line. It can be one line for entering long and another for short like in workshop 5. And the valley/peaks and crosslines can be combined so that the trade is entered once a line is crossed and it exits once a peak/valley is found.
So I am not sure what is the best way to evaluate all those algos in the fast possible way.

Yes, it's possible to train workshop 6 one asset at a time.

Entering trades at peaks or thresholds of the same oscillator can produce very different systems that react on very different market inefficiencies. So it's best to check beforehand which inefficiency you want to exploit and which sort of trigger it requires.