Forum Replies Created
-
Posts
-
Hi Petko,
I’ve just done a bunch of work putting together a system for trading live. Its not quite there yet but I’d love your input and if you agree with my approach and direction.
Basically what I have done is generate a pool of EA’s on the H1 timeframe with a reactor end date early last year for 6 currency pairs (About 200-300 EAs for each).
Then I have done the following:
Step 1 – Run validator (/recalculated) to filter out only strategies with profit factor > 1.1 and R/D > 0.5 for both 1 month and 2 weeks pre trading data.
Step 2 – Selected the top 5 simply according to max net profit for the EAs to trade for the next month by adding them to the portfolio.
Step 3 – Recalculated the portfolio for the mock trading month and recorded results in a spreadsheet
I have then done this in a monthly walk forward fashion as shown in the table below (Each row shows the months mock trading results from the portfolio):
For the in-sample dates I essentially collected data for 6 pairs and with a few different filter types (I.e 3 months and 1 month, same dates but with R/D > 1 etc)
Then after reviewing the results I settled on the combination of the filter criteria and combining the three pairs above. I then used the OOS months to just run the chosen pair combinations and filters and recorded the results. CERTAINLY a lot of work behind that table which is a very brief summary.
Regarding the reactor settings, everything in line with your courses a few slight deviations with the most impactful being more bars (I used 10 years or data) and a lower trade count since I was a bit stricter with monte carlo.
Anyway, a few questions:
1. You will notice my pairs are slightly unbalanced with 2 EUR and JPY and 1 GBP and 1 USD. This just showed to combine best over this date range with this particular pool of robots along with the filter criteria I set. I currently tend to accept non perfectly balance pairings as long as they are not overly unbalanced? I did have EURUSD but EURGBP just seemed to have better performing robots so I decided to use it instead, would you agree that is sensible?
2. This seems a decent enough start to a system to run for a month but definitely needs padding out as its not making enough trades and in fact makes zero trades one month of the OOS.
3. To move forward and pad it out further I was thinking of repeating the exercise but with M30 and M15 timeframes to add another few pairs and diversity to the portfolio. Rerunning the reactor with different settings on the H1 timeframe for the pairs that didn’t give me particularly good results. Projecting forward I can see that this MAY end up with something looking like the following:
5 EA’s for the month of EURGBP, EURJPY and USDJPY on H1
5 EA’s for the month on EURUSD and GBPJPY on M30
5 EA’s for the month on GPBUSD on M15
I used to try really hard to aim for perfectly balanced pairs with same number of EAs on same timeframe but I am finding that sometimes I just get better performing strategies on different time frames and I’m currently just feeling like the above would be nicely diversified (different time frames, reactor settings) if that’s how it comes out, any thoughts?
Also if you are wondering how I calculated the portfolio combined drawdown on three different pairs, I just summed the max drawdowns of each and multiplied it by 0.6 as a very rough estimate.
Also in the table above Walk Forward 9 was last month, I intend to create another table with the in-sample up until the most recent date and then use that to trade on a demo account and see how the results come out in the next week or so, will happily post the results.
I know these are EA studio results on open prices so there is some scope with spread and inner bar but I have also spot tested many of the portfolio EA’s both with open prices and control points in MT4 with a spread of 20 and have been happy enough with the comparison. Demolive trading is the true test of course.
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
I’ve got my sights on passing the FTMO challenge, I’ll post here when I have a crack at it possibly in the next month or two depending on how the RussiaUkraine situation unfolds.
Hi Petko,
Yeah its more a good R/D (>1) I am aiming for rather than just a small drawdown. I’m aiming to have a go at some trading challenges like FTMO, myforexfunds etc and they all require about a 10% profit with a maximum 10% drawdown in a month so I figure if I can keep my R/D > 1 then have a decent shot of passing.
Sounds like we agree on the diversity of assets and strategies being the best way of controlling drawdown. I do include it when I am filtering strategies but I feel a well put together portfolio would have a much bigger effect on controlling drawdown.
Its also a shame the EA studio cant create a combined portfolio of different assets so can see the true combined drawdown.
Currently I just sum the drawdowns for all the pair portfolios and then multiply by a factor of 0.6 as an estimate, assuming that its reasonably conservative that most will be on their worst drawdown at the exact same time. I’m sure in many cases it would be a lot lower than this though but its the best I have come up with for the time being though.
Jaroslav, it definitely sounds like you are not ready to be moving anything to live yet. Use a second demo as a mock live account to move your EAs till, then once you have build confidence in what you are doing you will feel much more comfortable moving to live.
One additional suggestion Mario, it sound like you put this together pretty quickly on what I am assuming is a recent end date of the sample?
I would suggest repeating what you did exactly but set the end date back six months or so. Then assuming you have a collection of say 50 strategies that you think look good add a sample of them (or all of them) to the portfolio and then recalculate for the most recent six months (Kind of running a second OOS manually).
That might give a quick indication on the likelihood of current your process resulting in curve fitting. Also you mention running the generator rather than the reactor which also adds to this concern.
Ordinarily I wouldn’t think spread or slippage would wreck things (provided a sensible approach is taken) but if you were trading every bar like this then I’d say spread, slippage and commission would be extremely likely to kill it.
As Petko said the major concern is the length of the trade and that you have traded every single bar of the entire set? I actually cant even comprehend how such a strategy would be created. Its really just opening a long or short trade at the start of EVERY single bar and then closing it at the end of the bar and immediately reopening another long or short position which seems pretty wild.
Hi Mario,
You asked how realistic the strategies are:
What robustness tests have you performed? How many bars? What timeframe are you using?
As to widening spread killing the strategy two suggestions if you are concerned about this then here are a few suggestions to check this out:
1. You can simply increase the spread from 10-20 for example and see how it affects your strategy
2. Run a monte carlo on spread only
These shouldn’t show much effect as generally the strategy will simply improve if you lower the fixed spread and and vice versa if you increase it.
A check I often perform is to change the trading session time in tools from 00:00 – 24:00 to 01:00 -24:00 as I have noticed that this can be quite a volatile time for spread and so it has a measurable negative impact I personally don’t take that as a good sign.
Of course the most accurate way will be to run it on a demo and see the results when trading real-time with variable spreads.
If your strategy passes all those check then I wouldn’t be concerned about spread (In my opinion anyway), but there are many more checks required to increase the probability of your strategy being a robust one. Its as much an art as a science I reckon.
It can be quite easy to get excellent results as you have shown but you have to be careful that you have not simply allowed EA studio to curve fit results.
Resolved: It wasnt the software mismatching. I was using premium data in EA Studio and my broker data in MT4. For a couple of currencys broker data was more different than I would have liked but I got things to match by importing broker data into EA studio. Probably wasnt that different but I was just doing some checks and wanted things closer, was being a little bit too pedantic I think.
I think I have half figured it out. Seems that when I look at the indicator chart in EA studio that the bars start at 41 and if I go back to the last day of the weekend then the trades match exactly. I think it may be something to do with EA studio requiring a certain number of bars to calculate the indicators for the signal. Im not toally certain what exactly is happening though.
To add to this, I have run the visual mode in MT4 and also inspected the indicator chart in EA studio.
The entry signal is the aligator teeth corssing above the jaws for the buy and opposite for sell. Everything matches up except that according to the indicator chart in EA studio a signal is clearly made but no entry is applied?
Hi Petko,
How often do you update the pool on the course? Sounds like you update your personal pool almost daily but I am assuming that the zip file on the course is updated monthly?
Hi Petko,
Yes you are correct that there is no issue, I could not replicate the problem.
I think I must have got myself mixed up between the data sources and jumped to the embarrassingly arrogant conclusion that the software was making a mistake rather than me.
Hugely appreciative of your content btw!! Its changed everything for the better for me
