Hi Team,

Just trying to put down a few thoughts on my approach to creating EA’s & portfolios from a top down approach.

I am a new trader and have seen endless opportunities and ideas to explore with ea studio but cannot stop thinking about it hahaha. I will put down some ideas here hoping to see what you all think and for the communities benefit! I will reference AUD/USD pair which I have been working on lately.

 

PLAN

1. Choose Pair with small spread. (AUD/USD)

2. Download historical data and upload to ea studio.

3. Set relaxed acceptance criteria (PF 1.2 & Min count of trades 300)

4. play with Generator settings to see what achieves the most strategies. (I personally like No TP and a trailing SL)(I also like using no exit rules apart from SL & TP)(Must play around a lot in this step to see what works for the pair on that timeframe)

5. Put the whole collection through the reactor and Monte Carlo (random backtest, spread, indicator, starting bar)(I always use these 4 Monte Carlo checks)

6. Now I have a collection of EA’s that have been through Monte Carlo. (I download and save this collection of 200 EA’s)

7. I filter the collection of EA’s by R2 and take the top 10, I then filter by win:loss and take the top 10, I then filter by net profit and take the top ten. (no EA is added twice so there may not be a total of 30 EA’s after the filtering so I manually add a few more until I reach a total portfolio size of 30 EA’s.

8. I now have a portfolio of 30 EA’s for a specific timeframe.

9. I repeat the above for all the currency timeframes. (I intend on only using the best timeframe for each currency)

10. I then take my original collection of EA’s and put them through optimisation and repeat step 7 of filtering via R2, WinLos & net balance.

11. I then take my original collection of EA’s and put them through walk forward and repeat step 7 of filtering via R2, WinLos & net balance.

12. I then take my original collection of EA’s and put them through normalisation and repeat step 7 of filtering via R2, WinLos & net balance.

13. so I know have 4 Portfolios to test on a demo account.

PORTFOLIO 1; 30 Filtered EA’s after generator and MonteCarlo.

PORTFOLIO 2; 30 Filtered EA’s after generator and MonteCarlo then optimised.

PORTFOLIO 3; 30 Filtered EA’s after generator and MonteCarlo then Normalisation.

PORTFOLIO 4; 30 Filtered EA’s after generator and MonteCarlo then Walk forward.

14. review results from the 4 portfolios.

 

Again, I am new here but can’t keep up with opportunities to explore! if anyone has any feedback on the above approach it is greatly appreciated.

 

Kind Regards,

Michael Richardson

 

Hey Michael,

Thanks for sharing your ideas!

All seems good to me, and here are a few tips:

2. You don;t need to import data to EA Studio – it’s data is briliant, just make it work for your broker by changing the symbol settings.

9. You don’t really need to do it for all timeframes. You will see that there are a few that works best.

Also, at the end you can recalcualte the EAs without using the last one month of data. So do all of than without the last month (you can check it from Data Horizon) and then uncheck it which will be a good forward testing.

Cheers,

Thanks Petko!

That’s a great idea, I will start using the EA Studio “Premium data” and adjust the symbol settings to suit my broker and focus on the best timeframe.

My process is continually changing at the moment but seems to be progressing well.  I have a few strategies playing out so will keep you updated 🙂

Regards,

Michael Richardson

 

I just started the free trial of EA Studio. I went through the course but for some reason just can’t seem to generate many strategies. I’m currently on EU M15, FTMO data.

Thank you

Hi Brian,

The first thing you can do is to just run the generator and not the reactor and only put profit factor of 1.2 in Acceptance Criteria.  You should be able to generate a bunch of strategies.  Once you get this working then you can advance from there.

Keep me posted!

Alan,