[Sticky] Lesson 3: How to prevent curve-fitting.
Good questions...yes indeed for many newbies trading using SQX will be very hard to conquest...many will give up after many many tries to make money with it. The course provided by the company is not enough to make money with this tools...it will just not happen for many of you.
So I guess SQX is only good for the diehards out there...newbies that never give up, or traders that already have a profitable trading plan and know what they are doing...
So the bottom line: SQX is a nice piece of software (but still somewhat buggy), so get ready for a very long ride before you will see any $ in your account.
I trade only futures on TradeStation, for the following reason:
1. All FX brokers I have seen so far are scammers to some level...so I prefer semi professional brokers like Tradestation
2. You will probably never get a good datasource from your FX broker, your SQX results will be always be different than your live results = lack of correlation
3. FX strategies are stagnating for the last 2 years...all FX portfolios I have seen so far go flat in best case!
If you need any more info just mail me,
I hope it helps
Hello Coensio, first of all thanks for you answer ....
I agree on the issues related to data, and unfortunately related to brokers too, but let's try to focus on the key factor: the ability of SQX to generate strategies worthy of being sources of income.
At first glance, I do not think the method of putting together blocks and indicators together and making a smoothie is effective (or only apparently), this is why my effort with SQX consists in the ability to "customize" the bricks ... representing a strategy I have in mind (with a logic), rather than relying on generations of strategies randomly combined by pieces of logic. what do you think ?
you say: "FX strategies have been stagnant for the past 2 years" ... can I possibly address this limitation by adopting a valid "robustness test" process?
in your opinion, what would be the key steps? applying Montecarlo on parameters followed a robust application of the WFA is enough ? then, do you believe that the application of rotational portfolios can refresh strategies in order to adapt to market developments?
Thank you for attention ! ...
with cordiality, Sabino.
I have in mind (with a logic), rather than relying on generations of strategies randomly combined by pieces of logic. what do you think ?
This approach works if you know which blocks to use for which market. Selecting all blocks and indicators and randomly generating systems takes a long time. And I'm still waiting for SQX team to release a SQX version where we will be able to code our own strategies with our own logic....
SQX is a nice tool but nothing more than that, it's up to you how you will use it..it's not a magical money making machine 😉
W.r.t FX-market results, there are many other trading groups like 'hankeys' group where members are collectively generating and incubating large quantities of strategies and trying to incubate them over a long period of time, selecting those with a nice backtest-to-live correlation and other good characteristics...that's one way to go, but it takes huge amount of work and time. And 90% of incubated strategies will be worthless in the end, not correlated to your SQX backtests and if you will find few good strategies it is still very very hard to make a profitable portfolio. What really helps is usage of recorded broker tick data...but still it will be really hard to make some profits
on portfolio level. Period.
The other way is to make sure you have a broker with a good execution and RELIABLE HISTORICAL DATA, where strategy generated live trades are very closely correlated (almost exactly the same) to SQX backtested trades. Broker example = Tradestation, markets = futures, mini/micro indices (Margin requires <$2000/system for micro contracts).
Then you can focus on real strategy development, because then you know that SQX generated equity curve that you see in your SQX, is very close to the curve you would get while trading on a real account. Only then you can focus on robustness and workflow filters etc...etc..to generate stable systems. Tip: concentrate on WFA systems and always leave at least 1 year or more of 'untouched' data to test if selected strategies from your workflow would really make you money. You can read my posts from forum, where I have described how to approach this.
I hope this helps...
whatever you write is help, thanks for your attention
"And I'm still waiting for SQX team to release a SQX version where we will be able to code our own strategies with our own logic...." 😲 😲 😲 .... really ?? is this functional extension foreseen?
how can i find the "hankey" group ? .... it's a private group ?
You can find his group on the official SQX forum 😉