Forum Replies Created
-
Posts
-
Hi Matthew,
The optimization settings seem fine to me, of course you can try experimenting with changing search search best Net balance to perhaps R/D or Maximum drawdown for example to help for the rules of FTMO challenge but for now if stick with what you have rather than changing too much at once.
Ah yeah, I have the same issue so yeah can’t check 1 week but can do 2 weeks as I mostly favour H1 timeframe also currently.
Additional detail with this is that EA Studio won’t make any trades in the first 100 bars of data, so if you want to check 2 weeks which is 240 bars for H1 then set the bars to 340 bars, you can check this is the journal and also will see the setting in the tools page. So if you put 300 bars in for H1 you are actually checking about 12 days.
One other suggestion is dont change everything at once but rather make a change and try to get a feel with some tests of how that single change has affected the results. Obviously more time consuming to start but in the long run I think you will win faster.
Hopefully your results start improving soon, let us know :-)
Ah hold on it IS a big deal that you are generating IS, definitely change that to be 30% OOS. You have changed the strategies by normalizing them but it definitely still takes away from the value of the OOS check considerably.
ONE last thing Matthew (Things keep popping into my head haha). Have a good read of some of the other forum threads even if they are a bit old, there’s lots of great information especially coming from Petko (It’s all here it just takes a bit of time and practice to assimilate)
‘what acceptance criteria do you use’ is a really good one for example.
One last point also, is that we are discussing obviously just one pair and timeframe and it could just be a bad month for that pair, so definitely a good idea to do some backtesting tests on the spare months day for say three well balanced pairs and them sum the results of the combined portfolio. Say EURGBP, EURJPY and GBPJPY for example.
It sounds like you are doing well to me btw, like you say theres perhaps just a few subtleties missing I think.
Plenty to chew threw from last post but small additional point is check what your acceptance is in the Monte Carlo, I am taking a complete guess that you are running indicators and backtesting date tests and that your acceptance criteria for the Monte Carlo is set to profit factor > 1??
If this is the case for example, I expect you might get a similar amount of strategies passing if you reduce the 90% to 80% and increase the profit factor to 1.1 or even 1.2 which may be worth a try, just a side suggestion/food for thought comment though :-)
Also my approach is generally similar and I have mixed results so I don’t claim to have everything figured out at all (but the harder I work the more things improve). I have been at it a while though and certainly tried and pondered lots for sure though so hopefully some of what I have said helps give you some ideas to move forward :-)
Hi Matthew,
It will be interesting when Petko responds which I am sure he will soon but here are some of my thoughts and suggestions.
First sticking with FTMO free trials is smart because you can do them as many times as you want with good metrix information but also just regular demos are obviously great too.
1. Why have you chosen to generate using the complete sample and then normalize only on the 30%, that is not logical to me (It’s not a huge deal but is there a reason?)
2. Your acceptance criteria looks fine to me although the complete pf>1.2 is redundant since you have it in both the IS and OOS but 1.2 for both is a good target in my opinion.
I totally understand the problem of the losing ones killing the overall portfolio, this is where running some tests will be really useful.
I don’t think there are any hard and fast rules as to how many trades etc, strategies generated are random in nature and so each sample of say 300 reactor results with same settings is gonna be a bit different.
Therefore, 1.2 for 10 trades may work great sometimes but not others. To counter this I would suggest holding a Month of data or so back when you run the reactor, then using once you have a collection of 300 test this filter criteria to see if it would have worked.
What will be easier than a count of trades to test will be a time period, so if for example you used the validator to see what would have happened if you traded only strategies that had a pf>1.2 for both a week and a month prior to your reactor end date then how would your results have been if you traded on the month of data that you held back?
<p style=”text-align: left;”>I hope this makes sense, your approach is close from what I can see. You are creating and robustness testing strategies to target a pf>1.2 and then only accepting ones that also stay true to this profit factor in the OOS, but then you are kind of randomly just selecting a filter criteria whereas I believe if you run some reasonably quick tests then you could give much more confidence to your filter criteria of over what periods you want to see a strategy stay true to this profit factor before you move it to live(or mock live demo in this case :-)</p>Also lots of other strong voices in here that haven’t been heard from in while (Andi, Haliffa, Roman, Jaquin2002).
Anyone about, how have your results been going?
Hey Richard,
Been reading through this thread and really resonated with your posts. Are you still on here? Sounds like you were honing in on success for sure to me so expect you are too busy enjoying your millions.
If not then it would be great to hear how you are getting on, especially with regards to getting stable results that stay within the strict guidelines of the prop challenges?
From what I have read my approach and use of EA Studio is very much aligned with yours so I really do hope we haven’t heard from you because you are too busy enjoying your profits!!
Hi Petko,
Thanks for taking the time to review this.
Yes I just used the reactor but it seems that the strategies are quite selective with their entry conditions and this can result in even a whole month of no trades at times.
It was primarily the process of walking forward through the months results and using an IS of months results to then decide how to trade the future few months that I was curious of your opinion so I am glad you agree that it is a sensible approach.
I’ll be doing a bit of work to run through this again but with some higher frequency strategies and aiming for around 6 pairs using a couple of different timeframes. I’ll keep you posted on how the results look :-)
Hey Petko,
I didn’t explain myself well there.
It’s fine that the portfolio EA is for one asset (which I agree is a MT4 limitation anyway).
More what I thought would be a good feature would be if EA studio could simply allow us to see the backtesting results for a combined portfolio of assets? That way we would get a better idea of how our portfolios on different assets are complimenting each other.
But of course we can only download a portfolio EA for a single asset and timeframe but that’s probably best for tracking purposes anyway.
I’m pretty sure it’s a feature that Miroslav could implement. It’s not critical, but would be nice.
Hope that makes sense. And I agree MT4 is the best in my opinion also
Hey Matthew,
Just to add to my previous post. Either by using demo or non demo filter methods, principally we are trying to do two things I think:
1 – Trade only the Strategies that are staying true to their generated acceptance criteria
2 – Catch these strategies in their profitable phase
So with your tests you definitely want to play around a bit but also keep this in mind to avoid being too random in your approach of course.
What broker are you using and have you uploaded their data to ea studio? What longer term filter are you using and is this definitely fully past the date range used to generate the strategies you are using(including OOS)?
Also you say you have put on Top 5 etc with poor results, presumably this is top 5 according to net profit. Perhaps may get better results if put on top 5 according to R/D or max drawdown?
Hi Matthew,
Profits do not come easy and Petko’s advice has always been to learn and practice using demos before using live. I made the same mistake early on but to a much smaller degree than 10k fortunately.
You must practice on demos until you are profitable.
Regardless of how skillfully a collection of robots has been put together if you just throw them on a live account you will very likely lose money.
As to the recommended time etc, have you run your own tests?
Countless ways to do this but you need to get creative, for example. IC markets allows up to 20 demo accounts, you could open 1 to put on all the robots you have then come up with 10 or so variations of moving robots and keep track of the results in a spreadsheet and see what is performing best as a filter criteria maybe?
For example
10 trades with pf>1.2, 5 trades with pf>1.2 and win/loss > 0.6 and so on.
Play around and learn, but definitely don’t go live until ready or you will lose money for sure regardless of the quality of the robots you start with!
Cheers Petko, that means a lot coming from my Teacher.
It wasn’t until I found your courses that my passion and enjoyment of trading really took off in a big way, been working hard at it ever since and loving it :-)
And yup, insane strategy that does show how powerful EA studio is, it’s fit something that looks awesome but I can’t see how it could have any predictive edge to it!
Reconsider the settings, just recently I thought I had some fantastic reactor settings going but realized that somehow they were resulting in a huge proportion of my strategies in the collection making most of their trades on Monday at 0:00 or 1:00 which is high spread period. Without going into it too much it wasn’t gonna work out well in live trading!
It really is absolutely brilliant software that does a huge amount of work for us but we still need to a good bit of thinking and run some sense checks and tests to deepen our understanding of the end result and increase the robustness of our strategies.
Practice makes perfect as they say
Hi Petko and Adnan, just thought I’d mention it is actually C# that fsbpro uses rather than C++.
I personally have a big aversion to C++ so I’d probably straight away think “nope” if I had to do anything in C++, but C# is MUCH nicer and easier to knock something together if have a little bit of general programming experience :-)
There is a large repository of indicators for FSBPro where you will be able to see the and copy the code, so I’d just copy the code from one of the simpler better written ones with clean well structured code (choose one that has been written by Popov) and make some modifications if you want to put a personal indicator together.
Hope that helps