Trading Naked EA

Hi everyone, I am new to forex and this is my 1st post in FF.

I have developed an EA solely based on price action without any indicators. So far I have backtested it on GBP/USD from July 04 to Jan 07 on Alpari MT4 only and the M1 data was downloaded from Alpari data bank.

Anyway, I am not sure whether you guys are interested because it cant double or triple up your money in short period. If you are interested please let me know your test result especially if you have an older databank.

The attached EA is the beta 1 version therefore I dont provide user input option at the moment. I prefer to set the variable in the coding in development stage so that it is easy for me to recall it for testing. And, please take note that this EA only perform backtesting on GBP/USD only.

Hopefully this is not another curve fitted EA…

 

As you can see in the snapshot, there is only 121 trades in 2.5 years. So such kind of opportunities hardly exist everyday or may be week.

Anyway, the best setting is not included in the beta 1. I still can boost another 50% extra profit without significant changes in other statistic number.

 

what is the purpose of your thread?

If you want to sell your expert then go to commercial section…

If you want people to help you improving the resutls then share the method and the source file

He is sharing the method. Perhaps he wishes to protect his source code. I know I would from Ebay thieves.

Although, you really should allow forward testing on this EA, Backtesting is only a part of the test.

Yes, you are right. I do not want people C&D (Copy & Develope) and then commercialize my hardwork without my permission.

Anyway, you still can do forward testing. I havent done any forward testing because I just finished the basic coding a few days ago. However, it may take about a year to have 50 foward tested results before we know whether this is a curve fitted EA. Therefore, I hope someone who has an older databank (unseen data which is out of my coverage) can help me to do backtest. If this EA cant archieve a good result by using older data then no point i continue the development, right?

Please bear in mind that this coding is still in preliminary stage (I am yet to name the EA) because the only exit strategy is either it hit SL/TP or close by the end of the day, and it doesnt have any money management strategy too. The basic element must work 1st before i further enhance the coding.

Ok the basic concept of my EA is very very simple. We enter with the same direction of breakout only if the price pullback from previous day high/low. However, this basic idea doesnt work alone because it will have a lot of noise.

The coding part that spend most of my time is filtering. Depending on the filter setting, it will create 3 type of equity curve pattern, which is consistently going up, going down or no direction.

Therefore I have created another set of coding which is totally reverse the original concept, instead of buy we sell. And this make the consistently going down equity curve turn into consistently going up.

In order words, the beta 1 has included the basic idea of entry and the reverse entry.

So far the pullback/SL/TP is fixed so that it is easy for me to do statistic analysis. The current setting in the beta 1 is 20/30/60.

I Believe your findings are valid. seeing you have a large sample in your backtest you could have found something here. and your entry being so simple you could manually backtest your strategy or use visual mode, and if it works then it works, Only thing left to do is forward test the idea.

Regards,
Mikhail

There are a couple things to look for which can be warning signs of curve fitting.
1. If you’ve got lots and lots of parameters and test over every possible combination. It doesn’t sound like this is the case for you.
2. A small number of trades entered or short backtest period. The less trades entered, the easier it is to curve fit to get sparkling results. I think 121 trades is a bit on the light side but doesn’t shout “danger” to me.

It’s a good idea to optimize over a period and then check your results on “out of sample” data. Example: optimize on Jan 1 2003 through Jan 1 2006 and then check results without further optimization for Jan 1 2006 to present.

Enclosed is the beta 1.1. I would appreciate it if I could have some feedback.

Changes:
1. Added user input option.
2. Added trailing stop feature.
3. No longer limited to GBP/USD only.
4. Added entry & exit arrow on the chart.
5. Pending order expiration changed to 3 hours before end of the daily bar.
6. Minor bugs fixed.

Recommended setting for GBP/USD(I havent tested on other pairs):
1. TP_pip : 60 to 180;
2. SL_pip : 20 to 30;
3. TS_pip : minimum 50;
4. SafetyMargin_pip : 18 to 22;

I hope the below picture can provide you a better visualization if you have no idea what is the thing behind the code.

Leave a Reply

Your email address will not be published. Required fields are marked *

ten + thirteen =