hello from Gianfranco i am new in algo trading..i send this little work ( iam not professional)..
Hi, as a new member you need to reattach your files after your first post...and make sure the files are zipped.
Thanks,
Chris
Good work 🙂
montecarlo test 1000 runs randomize strategy parameter 20/30
and wfm 5000 runs 0.1 lot size or try fixed amount risk 100 usd
good night
Gianfranco
Hello, I loaded it in strategyquantx. It saids failed
thanks, my bad. I just noticed that he had a different set of "WF conditions" .
The test finally finished. I ran it thru my WFM, I used default settings from 2005 - 2018. It passes at 67%. I used automatic settings for the parameters.
My question is, what is best to use for parameters? Should we limit what is included (can choose what to include on the left: constant, periods, boolean, etc.)? Should we use manual or is automatic okay?
Because it seems maybe the parameters are best limited to indicator periods and SL & TP. Maybe it might change the accuracy of a WFM? Does anybody have any experience and/or views on this?
stearno
I think the best way would be to select those parameters manually, (I would skip boolean and constant params), for example you could do a very wide parameter (normal) optimization at first, then use 3D graph to see which parameters really matter and what would be the best optimization range. Then you could use those parameters + parameter ranges to do your WFM. Of course keeping in mind that the number of combinations could explode and you need to do WFM with at least few % of the total optimization range...
So it is not easy and very time consuming...in most cases I also use automatic range (+/-30% or 40% of parameter range) Max optimizations = at least few % of a total range, but I do this only for additional strategy robustness confirmation not for actual trading with optimized parameters. If you really want to use WFM for trading then one simple solution is : keep your strategies very simple (= minimal amount of parameters), which also helps to prevent curve-fitting.
Hello by Gianfranco...another strategy...eurusd h1
@Gianfranco it looks good..can you also add the actual type of MC test so we know which test is which one?
Thanks!
hello from milano...mc test 500 runs randomize trades exact...skip trades 10%...(randomize strategy parameter 20 %probability...max change 30%)
hello Gianfranco….a folder with good strategies …….1 of my fortfolio
Thanks for contributing, I 'm currently on my holidays but when I'm back (in two weeks) I will filter your strategies out and see what you have there...
You know that another very good 'technical' pairs are: XAUUSD, EURJPY, GBPUSD and AUDCAD maybe you could try to find some good strategies on those pairs to diversify more in your portfolio.
Gr
Chris