Hi all,
Just released one of my most promising expert advisers as open-source code. Download .ex4 and source code at tradeharmonics.com/GoldenEA.zip
It’s not perfect, but back-tests very well and live demo testing has gone well so far. Can’t wait to see what some more talented minds than mine can do with the program.
Cheers
So far testing has been ok with this EA. 8 small wins (5 pips), 1 loss (-30 pips).
that’s not a very good risk/reward ratio.
I decided to try it out. But for some reason it didn’t make any trades back testing. Any idea why?
SL=30 and TP=5 are the default settings that have been highly profitable over a 10 year time-frame, but are not currently making profits in today’s volatile market conditions. The combination of a simple Golden Ratio-based Fib bounce and a simple Golden Ratio-based money management (5/30=0.167) was extremely effective during times of stabler market conditions, but is too tight for today’s volatility.
The program needs to be optimized to be profitable in the short-term. Check out the back-tests online and in the download package for settings that are more likely to be profitable in current markets, or run your own optimization tests based on a profit-curve fitting strategy that you believe in. Optimized settings between 2007 and 2009, although allowing for much more significant price movement, would also be considered to have more acceptable risk/reward ratios. (Optimized at SL=85, TP=270 for 2007-2009 on the USD/CHF 5 Min). While this fundamentally changes the strategy of the program from a scalper to a more long-term strategy, it is nonetheless an effective method of avoiding loss, and hopefully making profits, while markets settle back into fundamental patterns.
TalonD - Not sure why the program is making no trades. There is a spread limit function that stops the program from trading when the spread is too big, but that shouldn’t be a problem on the USD/CHF or EUR/USD. The system has also not been tested on terminal software that uses 5 decimal pricing, so if that is the case with your historical data that could be the issue. Default settings are also to trade micro 0.1 lots. This shouldn’t matter for historical testing, but could potentially cause problems in live testing on standard accounts with some brokers.
Good luck, and thanks for taking an interest.
Interesting, I was involved in a project where we made an EA with a 91% hit rate out of 100s of trades, TP4 SL30, and it wasn’t that amazing after all the effort simply due to a low profit factor
Could you provide any more info on expected results?
Sim - Check out the back-tests online at the Trade Harmonics website, or download GoldenEA from the download link in the first post. There is a bunch of testing reports and optimization data included in the download package. Plus, it’s free and the file is only 600k, so you download it and run your own tests to your heart’s delight and decide whether or not it is a program you want to use on a live account. So far, most promising results have been from testing at default parameters from 1999 to 2009 on USDCHF 5 Min or 1 Min.
noob question, how do you load this into MT4?
aaron3719 - You need to copy the GoldenEA.ex4 and GoldenEA.mq4 files into the /experts directory of your Metatrader program. Read the manual that is included in the download package, it should give you enough information to get started. Take note though, if you are unfamiliar with the use of Metatrader and EA programs you should probably educate yourself a bit before using GoldenEA. Like any trading program, if used incorrectly it may incur losses. Good luck.
ok thank you!
Oh damn, did not see your post previous to my reply, sorry!
Yep will download and test as soon as my net recovers from this crazy lag tonight, and will give feedback soon. Seems promising - you know your stuff!
GoldenEA,
Thanks for sharing the EA. 1 question though; what time frame the EA is designed or good for?
Thanks again.
Soul
At its default settings SL=30 TP=5, the program is most profitable on smaller time frames. The 5 min is ideal for testing purposes and provides the best mix of modeling quality and profit. Higher profit is generated when testing on 1 min timeframe, but modeling quality cannot exceed 25%. Either way, default settings are not currently profitable.
TP and SL settings can be optimized (see above replies) to be currently profitable on the 5 min timeframe. Optimized parameters for current market conditions, however, change the nature of the money management system from a scalper to holding longer-term positions. GoldenEA can currently be optimized to produce profitable results on higher time frames, but users with the experience necessary to make the program profitable on higher timeframes should be well-familiar with the Metatrader strategy tester and how to perform profit-curve fitting optimizations (a fancy way of saying optimize the program over increasingly smaller periods of time in order to fit the settings in the most profitable way to current conditions).
Hopefully I will have some optimization tests for different timeframes up on the Trade Harmonics forum in the next couple of days. Otherwise, if you are already familiar with optimization reports from the strategy tester, you can start be taking a look at the optimization results included in the download package.
I’m getting the same thing as Talon, and I’m using 5 decimals. When I check the strategy tester’s Journal, it’s giving me a [I]“TestGenerator: unmatched data error (volume limit 948 at 2009.08.17 08:05 exceeded)” [/I] and also [I]“TestGenerator: unmatched data error (high value 1.4543 at 2009.10.02 07:00 is not reached from the least timeframe, high price 1.4533 mismatches)”[/I]
. 
I use a 5 decimal broker for demo also, IBFX so that could be my problem. I don’t use EAs but the backtest results and concept of this one sounded interesting.
another interesting one is the nanningbob ea, but that’s off topic
by the way SP, I still havn’t read that book, it’s on my to do list with so much else.
LOL…I figured as much…I have one of those too…so much to do, so little time 
Thanks for informing me about the 5-decimal problem. I will get on fixing that and post an update here when the code is ready.
Actually the EA works in forward testing with 5 decimals…just doesn’t want to work when back testing in the strategy tester.
Here is a version with revised code for use with brokers that use 5-decimal pricing, such as FXDD. There is, however, one important change users need to be aware of.
SL and TP pip settings need to be multiplied by 10 in this version.
This is because 5-decimal brokers count their point value in pipettes instead of pips. Some other code has also been modified to make the program work better with 5-decimals. The regular download will still not make trades even if you multiply the inputs.
Currently, this version back-tests relatively well with IBFX. It shows similar consistent returns with default values (SL=300, TP=50… remember these are multiplied by 10 so they equal 30 and 5 pips of value respectively) up until 2006-2007, but suffers more drawdown (30% vs. 10%) than on FXDD over the 2007-2009 period.
An xls worksheet is included in the zip that shows some basic optimization tests results for 2009 and some more currently-profitable SL and TP values.
Not making this an official version until I have a chance to optimize it and improve long-term performance to be on par with its performance on 4-decimal brokers… but this gives people using 5-decimal brokers a chance to play with a working version of GoldenEA without having to wait.
If the normal version is making trades live on 5-decimals, it may work properly if TP and SL values are multiplied by 10, but you’re probably still better off using the version above. More info to come as I have a chance to do more testing on different brokers. Thanks for updating the thread with results for your own tests.
Optimizations attached for beta 5-Dec version on IBFX, USDCHF [B]1 Hr[/B] Jun 2009 to Present -
Best results:
Profit SL TP
18569.32 Stop_Loss=200 Take_Profit=2900
16952.07 Stop_Loss=1900 Take_Profit=1600
16437.45 Stop_Loss=100 Take_Profit=3100
15340.9 Stop_Loss=2100 Take_Profit=300