Have anyone managed to make a continiously profitable ea? - page 5
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
If a manual trader cannot explain what he's doing ..... he doesn't know what the hell he's doing. Enough said :) Imo
what do you use for storing the XML data on arrays? MQL? How?
I wrote an excel macro to download the XML, read it and output it as CSV for MT4 to read.... but there are also indicators that do that process inside MT4 itself.. I just figured it's not worth the extra effort if I want to overview the news myself every Sunday anyway. I'd be happy share the Excel file, especially if you have similar time-saving things to offer.... currently I hate how much time I spend programming, especially when these kinds of things are pretty much basic and essential for any EA with a realistic chance of profit... I'd have thought there would be shared versions available.
When MT4 opens the CSV it reads line by line scanning for relevant currency/text/impact/whatever and then saves the data in arrays: datetime array for the dates, string array for the texts etc., with the indices matching. So you know event [i] is called text[i] occurring at event_date[i] with currency cur[i].. (but remember many currencies affect many others, e.g. CNY news will likely affect AUDJPY quite strongly).
@ydrol, completely agree! It's simple enough to tell a human, e.g.: "don't trade at news events". But what does that mean for the programmer...:
1) download news events.
2) parse news events.
3) stop EA from opening trades "near" the news time
4) perhaps make sure EA closes current trades with more urgency as the news time gets nearer
5) choose which news events will not affect trades so much (e.g. CAD news might not affect SGD/JPY)
6) if the news data proved to be unreliable.. find a new source and repeat
I think number 4 is maybe the hardest to model a real human's response. Again, I think many programmers will rush the coding for tasks that are 'simple' for a human, because they don't fully realise they are coding the equivalent of our subconscious brain, and therefore get frustrated when it takes longer than expected to get it right.
I know people here have tried neural nets in relation to trading and say the results are not great. But I what if they used neural nets purely for those parts of the EA that struggle with strict rules (e.g. point 4 above). I mean, a beginner trader would make mistakes closing trades too early or too late before the news.. but an experienced trade would make fewer mistakes.. there's definite learning from experience going on there.
What do you think?
Requirements development and capture is one of the key deliverables when automating a manual strategy. It's a real skill.
There are a number of stages.
Coder - What do you want?
Trader - Here's what I want.
Coder - This is how I understand what you've asked for, is it right?
Both - Iteration around the above until every tiny detail has been fully described and fully, mutually understood. This is both in terms of the strategy's functionality but also the technical environment within which the bot will operate.
I have no experience of the 'jobs' area on the MT5 website but I would guess that the above process is totally underestimated and under utilised by many coders and traders and thus becomes a buggers muddle on a regular basis.
yes, and then once the simple requirement is understood (in terms of a full implementation of a manual strategy) the coder has to either:
1. convey to the trader the complexity of a robust complete solution (to justify the price), or
2. do a simple implementation, which will likely be either buggy, have gaps, or be suboptimal.
I would think a continuously profitable EA needs a lot of support libraries (News, Holidays, Timezones, Trendlines, Support/Resistance), and would use multiple timeframes to determine entry setups etc. and have robust error handling.
If someone has written a simple EA (eg less than 2000 lines of clear code) that is continuously profitable I'd be both impressed, and inspired!
1. convey to the trader the complexity of a robust complete solution (to justify the price), or
2. do a simple implementation, which will likely be either buggy, have gaps, or be suboptimal.
If someone has written a simple EA (eg less than 2000 lines of clear code) that is continuously profitable I'd be both impressed, and inspired!
The problem would be finding a continuously profitable strategy which could guarantee its continued profitability. Since this doesn't exist (imo), one would be better off using simple strategies which he understands. News trading falls in the category of fundamental analysis and_not technical analysis. Expert Advisors falls in the category of technical analysis. There's obviously some limitations using Experts for fundamental trading. Like suggested, for news, allow most of the parameters as external and that'll save you allot of development and support.
Whats the point when the 200,000 lines generally doesn't work much better than the 2000 lines?
There is still often a requirement to avoid news and holidays with technical analysis, (even straddling the news announcements is probably more technical than fundamental?) so you still need to know when it is, but as pointed out it may be easier to give the trader a switch.
Regarding code length the point I'm making is that some simple things that are fundamental to technical analysis - like trendlines and Support/resistance - often require a fair bit of code. This may be tucked away in an indicator, but many 'simple' EAs seem to ignore these and concentrate on the statistical indicators rather than the price action based ones.... Again the counterpoint is to get the trader to draw the SR lines, and have the EA use them...
A number of EAs at Birts review have been making 1.3% per month for over 50 weeks. Thats a nice target (~ 16% PA compounded?).
I dont mind having to retune an EA every year or so.
So, to recap: one or two people have "sort of" made profitable EAs; one person said unreservedly that he/she has, but his/her testimony was thrown out due to a signature link; a self-ascribed successful hedge fund manager of over ten years says it's possible, but with caveats; a forum veteran has not yet been able to achieve it; yet, there are bots available for purchase that are independently reported to be successful. Am I correct?