Forum Replies Created
This change affects not only Fibo but other stuff too. Don’t know exactly what but I’ve seen huges changes in some strategies and they didn’t have Fibo. It’s a mess because as Mabi said I also have thousands of strategies backtested with good results and now all of a sudden I don’t know which ones are affected. I don’t see other solution that retest all again and throw away those that were good with B122 and with B123 don’t pass my tests. What a mess gentlemen
Lucky you! I’ve stopped all my strategies on EURJPY except one that is killing me
Attachments:You must be logged in to view attached files.
Hi Mark and dev team, to be honest is hard to me to understand how is it possible to freeze the work on fixing SQX when there’s a huge list of hundreds (if not more than a thousand I didn’t bother to count them) of bugs – some of these bugs are critical; like losing bars on the clone process in DataManager.
Personally I gave up months ago to use any new chic feature until that wasn’t completely tested and checked. I just go to the basics and expect SQX to work as SQ3 did. Would be great if SQX were faster but it isn’t. Same configuration on both and I get same strategies number on the building process although yes the strategies per hour rate is much bigger in SQX indeed but this is worthless if I get same number of strats (same configuration so the results are comparable).
When you were trying to launch new Builds each two or 3 weeks. I always said that it was better you could do your job and not deliver anything until was completely tested and if that takes 1 or 2 months, better take your time and deliver a quality work not anything that does whatever and generates a new surge of bugs and more problems and frustration.
Well, it’s the way it is. I would just settle to have SQX working same as SQ3 (not even better) and specially having SQX relaible on the basic stuff. For that reason I humbling ask you, could you focus on fixing the critical problems?
Well, maybe we’re talking about obvious thongs but if @notch you say Genetic algorithm is not searching by improvement… well nothing to argue, look at wikipedia.
And yes, curve fitting is worst nightmare. All of us are curvefitting, if not look at this sample of 3.000 strategies. They have an average Profit Factor = 1,6 aprox on 15 years. After the first 18 months this 3.000 strategies go to PF=1,4. That’s Mr curvefitting in actionAttachments in this forum are visible only to Customers.
Hey, debate is nice :)
I’ve checked both strategies; Marcel and hankeys – thank you both for your generosity sharing.
Hankeys looks nicer – better Ret/DD – but from my point of view is clearly less robust. The MC tests are failing following my personal workflow; MC Exact, MC Random. So, always from my perspective I’d choose Marcel’s.
Regarding the strategies improvement I don’t like it. Strategy is improved enough with the genetic algorithim, in our business Improving means curve fitting and that’s worst nightmare