Build SQX or old SQ3 strategies

Hi!

Is it bad if I build strategies with SQX but used the old SQ3 style? What are the disadvantages at old SQ3 style?

Brg: Csaba

https://strategyquant.com/doc/strategyquant/strategy-style/

Yes, thank you Hankeys.
I don’t know, why the old SQ3 style would be really worse. And I have the feeling, that with old SQ3 style the generation / number of succesful strategies are better.

Or am I alone with this feeling? 🙂

i am not focusing only SQX type of strategies, i am generating also SQ3 type of them and i know that they are not correlated, so i am getting more diversified portfolio

what is faster? i cant say, it could depend on market, etc.

Under “faster” I mean, that I have the feeling, that let say from 1000 generated strategies I will have more, they fulfill my expextations on long historical data. So I will get more nice equities, with which I can mae the robustness tests…

How do you mean, that you are generating also SQ3 style “of them“? Let say, you see, that there is a good SQX strategy with 3 indicators and than you will generate strategies with these 3 indicators, but in SQ3 style?

no i am using all building blocks (only volume and bar minute not), i want that genetic engine will take care of it – i dont know what will work…i am an algotrader and have app for this

so for example i am running SQX project and after 2 days i will run the same project but with SQ3 style

In this case we are on the same opinion. I just let run 2 weeks for each method.

I noticed one thing: at SQX method, there is very much times the condition “XXXX is rising for —- bars“. At the start I had everything switched on, but this conditions is very-very often there.

And one more thing: the Daily LOW, HIGH, CLOSE, OPEN indicators: if there was generated a strategy with these indicators, the source code of MT4 writes, that there is a known problem at backtesting in MT4 with these indicators, but there is no exact solution for that. What is the problem?

Hello,

are you talking about specific strategy? What kind of problem is that with OHLC prices?

Hello Tomas

No, it is a gereal info. Please see the attached picture and strategy.

Brg: Csaba

Csaba wrote:

….. let say from 1000 generated strategies ….

during my time with SQ3, i can confirm to you that on most pairs and timeframes, 1000 strategies in databank is not suficient to find a strategy, in most cases i used 5000 with modest results, so the number should be more close to 10.000 which would require a very powerful system or a lot of time

i must note that i noticed significant differences between the final group of strategies left after all steps of selection were over. In theory, if one could put on test all strategies, maybe one could find a working one but from each group, you are limited on how many you can put on test