Reply To: EURUSD M5 settings

#4600
DominikDominik
Administrator

@DimaExtreme wrote:

Hello!
I am trying to optimize eurusd setting a pair, but it does run on the adviser. Then he looked, that you put your sets and downloading the set, tested it on more accurate quotes. Here is a look at the result of his set with a fine sweep to 99.9%.

Hi!

I don’t understand in 100% what you wrote, but I think that I can guess most of it :) I would be very careful in using the system on real account with settings optimized only within two years. You know what I mean? Even if it may be potentially more profitable than optimized for 10 or 15 years – it’s riskier. My approach to this is that the system should be the most stable and resistant to different market behaviours. I compared results of LIVE trading with different settings. Settings optimized only for the year 2016 brought more losses than gains. After many small profitable trades (95% of all was profitable), there was only 1 loss, which took 70-90% of previous profits and it’s not good. I would like to focus on creating optimal settings that will work for a long time in future, so I’m attaching for you to this post best parameters that I have now for EURUSD M5 2000-now.

Below are results of optimization:

As you can see, I changed setting of dynamic lot size from 20 to 10 (that means for each 1000$ on account it will open transaction equal to 0,10 lot). I did so, because I want to trade parallel on other currency pairs to have diversification, greater growth of profit and several opened orders each day. Now, on the safest parameters (over 97% profitable trades) optimized for over 16 years for EURUSD on M5 TF system opens average 1-2 orders each week. Trading only one pair profit will be very small, but when we will add 10 or even 20 another pairs with optimized parameters it will look quite well.

For now, Expert Advisor doesn’t have in code news filter, so when someone want to try this EA on live account (as I’m doing) I suggest to turn it off before news with high impact are published. As soon as I find some time I will certainly add news filter to the system.

OK. Let’s say that parameters are finally as good as possible and now it should be checked last one thing on results tab in strategy tester: when the system was losing money and why? This would eliminate last losses. And there is still only one question, what can I change in system to have better results – higher profit in shorter time? Which condition or what in algorithm can I improve/change/delete? I tried different methods, but as one proverb says: “Two heads are better than one”. So, I’m waiting for every idea, no matter how simple or complicated it will be. If you have any – don’t hesitate, just write, I’ll code it and we will see :)