Hi Petko / all

I wonder if anyone can offer me some advice. I have run EA Studio and generated multiple EAs that have the following criteria: 2 entry (one is time of the day and then one other), no exit, instead a trade would either hit a 30 pip TP or a 30 SL. The idea of course is to find EAs with a 50% + win rate to be profitable, or actually 51.2% after taking into account my broker fees. I have generated the EAs on 5m charts only, and for 6 pairs, EURUSD, GBPUSD, EURGBP, USDCHF, AUDUSD and USDCAD. I chose these for the low spreads, I’d rather my EA had to get to a 30.1 pip TP and 29.9 pip SL than 31 vs 29.

So I have 6 demo accounts running 99 EAs for each pair, 594 EAs in total, all now been running for 11 weeks. My plan is to take the best 99 and load then onto a live account from beginning of January, with low lot values. I hope then that their collective win rate over the medium to long term will be 55%+. Logic would suggest I amalgamate the performance from all 594 into one single spreadsheet (I use FX Blue so have performance spreadsheets for each demo account, each week) and select the best 99 based on win rate. But is this the best approach? I don’t feel very knowledgeable on risk management, should i be choosing an equal number (16 or 17) for each pair? Are some of these pairs closely correlated, thus meaning I should avoid too chose too many from correlating pairs? Or should I be check the entry criteria and avoid selecting EAs that share the same single entry criteria irrespective of pair?

Petko, how would you approach it if I presented you with my data from the 6 accounts and asked you to choose the best 99 to launch into a live account!?

Keen to hear any thoughts and feedback. Thank you.

Hey Steve,

Interesting approach from you. Thanks for sharing it with everyone.

Here is what I think but not take it as a piece of advice:

1. I wouldn’t go for 99 EAs on a live account (it’s a bit too much). Instead, I would select Top 10 from each or even top 5 from each.

2. I never found decent results on M5, so I prefer M15 and H1 the most. However, give it a try and see if it will work or not

3. I would create one Demo account that would be simulating the live account and for at least 3 months I would test the whole system.

Anyway, I like your idea and I think the test you are doing is pretty decent and it will bring you results.

Hey Steve,

Interesting approach from you. Thanks for sharing it with everyone.

Here is what I think but not take it as a piece of advice:

1. I wouldn’t go for 99 EAs on a live account (it’s a bit too much). Instead, I would select Top 10 from each or even top 5 from each.

2. I never found decent results on M5, so I prefer M15 and H1 the most. However, give it a try and see if it will work or not

3. I would create one Demo account that would be simulating the live account and for at least 3 months I would test the whole system.

Anyway, I like your idea and I think the test you are doing is pretty decent and it will bring you results.

Thanks Petko, I appreciate your advice. Maybe I will run the whole experiment again on M15 and H1 aswell!

When you say “choose the best 10 or the best 5 from each”, would you simply do this on win rate, would you be looking to rule out any correlating pairs, or EAs that work off the same indicators, or am i overthinking it? Should it simply be a case of choosing the best 5 or 10 from their win rates?

For me personally, the profit at the end is what matters the most.

However, if I am testing 2 Demo accounts, for example, one with EURUSD and another with GBPJPY, and see 10 EAs from EURUSD and just 5 from GBPJPY to meet my criteria, I would place 5 from each so I will keep the exposure equal.