Adaptable EA's

Well first off, hello everyone! I am fairly new to this site, fairly new to Forex trading, but have a good history in programming.

With that, here is my take on the whole EA’s loose money argument.

It seems from the reviews and testing of various EA’s, on both MT4 and the new MT5, that I have done…that most EA’s are only effective in very particular situations.

And even these particular situations can vary from user to user, based on the settings they are using for that EA.

Through forums like this, users post their current set files to share their success with other users, but these settings can be outdated when stumbled upon by other users. Or are not used at all, from users hoping their EA will make a turnaround.

So why not make optimized settings available to all users?

If you are using an EA that is asking you for settings, but you do not know what they do, how do you know how to set them? Or if you do not understand how an indicator is being used in the calculation of a possible buy or sell situation, again, how do you know what to set them at?

Here is my proposal that I am working on.

Metatrader allows you to read and write files. In these files you can place different types of information. You can also read this information by an EA.

A database can work as a repository for optimized settings of an EA. These settings can be tailored to each currency and time frame that users are using.

You can connect to the database, download the most current settings for your currency, and create a file that the EA can read for its settings.

You then put that connection in a program, give it a set update interval and now you only have to worry about updating settings for the trend of the current market. While your EA works on the trading for you.

What I have done so far.

–Created a Database that lists users, currencies they are subscribed to, and currency settings.

–Created a basic EA that uses dual trix, adx and rsi to calculate buy and sell signals.
–Created a program that logs into the database, downloads most current Currency settings for the EA, creates a binary file to be read by the EA at selected intervals (30 min, 1 hr, 6 hr, Daily).

Currently Testing on USDJPY 30 min and GBPUSD 30 min.

I will post results after there is history of trading.

Now, tell me what you think?

This is the first and only trade performed by my EA so far. It was on GBPUSD 30 min. Nothing so far on JPYUSD.

Impressive result :slight_smile:

Funny how you call it impressive when the file can’t be opened…

Is anyone else having trouble opening the file I attached in the 2nd post?

That’s interesting! I’ve thought of something like this…but using EAs themselves to merely look at the trends and variances of the prices. You’d enter a symbol…or heck, let it look at all of them…and it logs trend info, variances (high variances mean use bigger stop loss, etc.). So before trading, you run a “look-see” EA, and then choose which currency pair, stop loss, trend vs. oscillator strategies, etc.

I’m still a newbie in this whole arena! What worries me is the complexity of having to connect to some database somewhere, import data and then decode it and all of that!!

Where would this database be? Could people who want to throw others off be able to enter numbers?

Well, here is what I have been building.

The database is hosted on a web server, so anyone with an internet connection should be able to connect to it.

There are different levels of access to this database (Admin and user). Users only have read access to the database, while admins have full access.

The program I am using to retrieve from the database, only uses the User Access role. This program does the user authentication, currencies subscribed to, stores the location of where to write the files to, and creates the files that are read by metatrader.

The EA can be applied to any Currency Pair, but will only do trading if there is a file created for that pair. A file that is only created if there is a subscription to that pair in the database.

I am the only user that has Admin access to the database. I am the only one that can add items and settings.


Currently in an open trade of AUDUSD , bought 0.5 lots at 0.9887. Currently showing $134 profit with -$6 swap. This is only the second trade the software has opened.

Nope, it opened fine for me. Its a PDF file. Maybe they need to update thier PDF reader.

Kaui Hawaii

How are you calculating these optimised settings for each currency pair what kind of criteria are you using ?

I am just using the Strategy Tester that is part of Metatrader 5. I run a first pass, testing all values of the indicators used along with take profit for the previous month.

I then sort the list by profit. Any calculations that have a large drawdown % ( greater then 8%) are considered non ideal. I try to find those that are 6% or lower.

Other values are also reviewed, to make sure they are using appropriate values. Like as I have mentioned in the first post, I am using a dual trix (triple exponential average) method. If the time frames for those two indicators comes back the same or very close to one another, I see that result as not ideal.

This first pass takes the computer about 7 hours.

The second pass is to calculate the stop loss. This pass takes only a few minutes. I am looking for a stop loss less then 75 pips. I know that some may see this as a high number, but that is what all this testing is for.

If I cannot get a positive result, with a realistic setting for the stop loss…I start the process over.

I do not allow the program to set the take profit and stop loss from the server, I feel those should be up to the users along with the lot size.

I hope this answers your question.

Then a second pass is run

Last trade is down by 55 Pips. I think this is a lesson to only do trading while the market is open.

Still on the positive side. +42 Pips.

To do:
Money management … allow users to select % of account invested.
Add time frame for trading… only trade sunday through friday. Close out all orders before market close on friday.

Any other suggesstions or questions out there?

Yes. You really need to take a look at some of the old webinars made by the guy who wrote the wave 59 charting application.

He attempted to design a system by doing something similar to you, i.e optimising parameters on historical data, and then walk forward testing. He reached some rather interesting conclusions :smiley:

In simple terms, he started by optimising parameters based on a 1 year sliding window, then walk forward testing. Not surprisingly it didnt work. He reduced the sliding window to 6 months, the results where worse, he reduced to 3 months, worse still, he reduced the time period to optimising based on the previous months data, a complete disaster. The shorter his optimisation period, the worse the result.

He concludes that as soon as a particular parameter set becomes an optimum (and therefore identifiable to practically all market participants), it stops working. However, fading the optimised parameter sets gave pretty good results (but IIRC varience was absolutely shocking).

His research is well worth looking at if your serious about trying to trade using this approach.

What this tells me is that for the system wave 59 was using, optimization was not successful.:wink:

Looking at the website for Wave 59, it appears they are using some pretty off the wall strategies like astrological alignment.

I will continue to keep testing, and improving my current EA to see if I can draw the same conclusion.:smiley:

For this particular research he was actually using a simple moving average cross.

I am not using SMA, though I look forward to seeing how my testing goes as compared to those of wave 59.

I don’t think EA Optimization brings much benefit, at the end of the day, you gotta react to what’s happening now, not what happened 3 months ago.