Finally! Simple working EA..

New comment
 

Well, I spend like one month trying to write an EA. I spend days and nights trying to find out out the correct combination of different indicators, but with very little or no luck. Then I read few post on this forum regarding generic algorithm, so I decided to create something very simple. I believe all the indicators are based on the past date therefore it is not possible to predict the future since its constantly changing.

So what if you just based EA on simple algorithm no matter what the trend or the market condition is? Then you have like 50/50 change to make a correct trade. Then you just do a small tweak to it and here we go:





What do you think? It trades static 0.1 lots all the time.. Started with 2k since 01/2008.

Any comments?

 

I have some:

- Drawdown>56%

- Profit factor = 1.46

I think it's enough

 
Im gonna try to improve the drawdown.. Basically the EA keeps trading all the time. It either buy or sell and it only exit the position to reverse it.. Quite simple, im almost surprised it kind of works..
 

Hi Jirimac

I'm not sure I agree with your 50/50 idea. If the spread is 1 pip you have to make 100% profit on your stake to break even so there is an inherent negative bias.

 
Yes Ruptor, you are absolutely right. But lets say that you hold the position a day, then the move is big enough to make the spread irrelevant...
 
If you think about all the things that can go wrong with a trade you will realise the odds are more like 90/10 against you. By going for bigger takes you have to have bigger stops and the take has to be bigger so you are more likely to hit the stop. There is nothing that favours a trade. The brokers want to hit your stops, if the platform crashes you can be left exposed, EAs have bugs, I can think of many things that have cost me but none that have helped an open trade. I think when you realise all the things against you then you can guard against them and start to make money. Well that's my two peneths worth.
 

I totally feel what youre saying. Ive been trading forex for over three years now. It takes very organized person to trade, to follow your rules, etc. And I have to say, most of the times I lost trade was because I panicked :).

I always try to follow some indicators but at the end its my "feeling" about the direction of the market. I had several EAs based on indicators but the performance is poor.

Sometimes the indicators are right, but sometimes totally wrong.. So what if there is no working indicator at all?

What if you just make a random decision to buy or to sell and go for if? What if you just buy when its sunny day and sell when its raining outside? What are the odds you gonna win or lose?

I kind of created this random decisions and here we go :)

There is no logic in my script but it seems to work..

 
jirimac:

Well, I spend like one month trying to write an EA. I spend days and nights trying to find out out the correct combination of different indicators, but with very little or no luck. Then I read few post on this forum regarding generic algorithm, so I decided to create something very simple. I believe all the indicators are based on the past date therefore it is not possible to predict the future since its constantly changing.

So what if you just based EA on simple algorithm no matter what the trend or the market condition is? Then you have like 50/50 change to make a correct trade. Then you just do a small tweak to it and here we go:





What do you think? It trades static 0.1 lots all the time.. Started with 2k since 01/2008.

Any comments?

Wow! That's a great discovery. What is the small tweak that you made?

 
jirimac:

.. Started with 2k since 01/2008.

Any comments?

Are you saying the chart shows 18 months of trading? The EA made essentially no profit in the first 9 months and has made all of its profit in the last 9 months. Is there some correlation that you can make to distinquish what is different in a broad way between those two 9-months periods?


Essentially, all we ever look for is a bias in the trading algorithm. Everything we do creates a bias, often the bias is little, but then sometimes it is big. Is there something you can do to alter the bias in the first 9 months?

Cheers

 

Hi Jirimac

One of the Neural network articles talks about a strategy using the base algorithm that has a 50/50 chance. It then goes onto improve the result using the NN to give it an edge. As usual I can't find the article. I thought this one was it but can't see the reference of 50/50. Still it talks about a basic trading system and improves its odds.

https://www.mql5.com/en/code/7917

I will be interested to see your further progress and if you need any help you know where I am.

New comment
Reason: