Here, take this profitable ea

[Matallica1]

@roar Thank you for the clarification about using this EA. Advice is well taken I will only use on EURUSD M5. Thanks again!

[AlphaOmega]

Thank you for the inspiration - Ea Roar

[thekidfromcagnes]

Hi I am testing the ea on a demo account but I receive a lot of fifo violations messages so the trades do not go through. I know we are supposed to use a netting account but is there a way to circumvent (eg close all trades in the chronological order)? thank you very much and congratulations again

[roar]

@thekidfromcagnes I dont trade fifo, but using the mt4 version should work (with some mods at least). You can copy the blocks from mt4 to mt5 in fxdreema, with two browser windows.

[roar]

Out of curiosity I backtested this EA again... And its still profitable.

I'm going to leave this thread here as a proof that EAs can actually make some profits and its not always (though it usually is) only because the optimization tool "forces" a nice graph but future performance crashes. This EA hasn't been optimized in 5 years, which is nice, as the system looks future-proof.

The 2020 crisis caused a significant drop to the equity, which is understandable because markets didnt quite behave normally back then. Maybe that could've been avoidable with some filter...

[makabasaking]

@roar hello bro.i just bumped into this project mate.i ran a test on it and i dont see it taking profit.how do i go about getting it to close trades.
thanks in advance

[roar]

@makabasaking you need a netting type account because that is how most mt5 accounts worked back then.

[makabasaking]

@roar so the system might not be able to work anymore with the current platform?anything that can be done to it to get it to use?

[roar]

@makabasaking it just needs some changes to the closing mechanism

[mel2424]

thanks for this

[makabasaking]

@roar happy new year buddy.can you kindly work on the closing mechanismfor that.i have tried playing around with it and i cant seem to be winning.its making huge losses on strategy tester....thanks

[Zereleth]

@roar Hey Roar, I have been trying to use this as a tutorial and I have been learning a lot about the way EA's are structured from this and the way that you use variables and constants throughout. It's a really impressive EA but I am having some difficulty with understanding how it functions, for example your variable 'med'.

I understand med is a calculated midpoint between the high and low but I can't see anywhere in the code where you use a modify variable to create this midpoint. The first mention of it in the blocks I could find was block 11 which checks that it's value is not 0. Is it possible to set a variables value outside of the EA blocks similar to a constant? For example that med=high+low/2?

This is one of the many questions I have on how this EA works but I don't want to overload you, thanks so much for posting this, has been a real help in improving my understanding, would love to see more tutorials like this.

Edit: So I found it in block 23, this is so intricate I guess it's what it takes to make a successful EA

[roar]

@zereleth I made the EA 6 years ago, so at this point you probably know the system better than I do, lol.

I really should remake it, half of the blocks could be combined in one simple custom code block.
Median price + RSI seems to be the winning combo here

[seb 0]

@roar
Very cool, I probably wouldn't have invested on it by the 2017-2018 forward result, but with more hindsight the EA seems to hold very well on the long run.
What criteria do you/did you use for optimization ?

[roar]

@seb-0 I dont remember, lol.

When I optimize, I use recovery factor as the parameter. Nowadays I dont optimize at all.

[seb 0]

@roar
Ok, I personally use a custom criterion which is : recoveryFactoer*tradesNumber^tradesWeight, tradesWeight (0,1).
Do you mean that the learning process is incorporated in your EA, so you don't need to use the tester optimization anymore ?

[roar]

@seb-0 well, I do optimize some times, but much less than before.

To my experience, with optimization you can make any EA look super good. Whats the point... the performance will not repeat in live trading.

So, if the EA doesnt look promising with random inputs backtest, I dont usually bother to force it look good with optimization.

Its 98% about the strategy itself, 2% about the input values.

[l'andorrà]

@roar Wise words!

[seb 0]

Well about "pattern rigid" EAs the parameters values impacts the results a lot. Let say a simple strategy that is : buy when rsi<30 and sell when rsi>70, can produce just by changing the parameters values : buy when -rsi<-70 and sell when -rsi>-30, which is the exact opposite and so both configurations can't be profitable simultaneously. Machine learning like EAs, could potentially remains profitable no matter the inputs parameters values, by inputs I mean feeding data. But then the parameters inherent to the structure of the learning process are of big importance because anyway the market follows accurate quantified rules. Finally I think a successful EA will always be better with parameters winning on training data, on test data, rather than with loosing parameters on training data, on testing data. Probably..

[sunny7day]

Can't get it profitable with default settings on eurusd M5 using everytick method.