Forex Factory (https://www.forexfactory.com/forum.php)
-   Platform Tech (https://www.forexfactory.com/forumdisplay.php?f=69)
-   -   Help Needed to Debug Existing EA with proven strategy (https://www.forexfactory.com/showthread.php?t=751438)

grenell May 12, 2018 8:55am | Post# 721

1 Attachment(s)
Hi
New to thread. Read from the beginning. New to back-testing, not new to trading. Thanks to Phillip, Swing Man, Nihilist and other contributors on a interesting thread.
I installed V5.2 on Demo GU and GJ this week as a forward test. Both pairs sustained 1 big loss each during the past week. See screen shot. I will TEST on DAX/GER30 with 1 lot this week (My minimum trade size with Ger30). The GER30 is the only instrument I trade live at the moment. The hours are restricted, I will trade it from Frankfurt open to London Close, not sure if the EA is intended for this instrument. I did do a few days this week as a test run and it functioned OK.
Regards
Rod
Click to Enlarge

Name: smForexRM 5-2.jpg
Size: 367 KB

Philip Lee May 12, 2018 9:03am | Post# 722

1 Attachment(s)
Hi guys,

I'm back. I have back tested V6.1 of Swingman's EA for the GBPJPY pair for the period 01.01.2017 to present day and here's the result:

Click to Enlarge

Name: failedV6-1.jpg
Size: 377 KB

It was doing quite well having more than doubled the capital in less than a year but it eventually blown the account.

Even if it had survived and completed the test, I would still be not happy due to the fact that the deadlock situations had occurred on more than 10 occasions with the longest lasted at least 3 weeks with 163 opened orders. Realistically, I don't think anyone of us could accept that in live trading.

I noticed that Swingman had implemented quite a few features into the EA but I must admit that it has little effect in reducing the deadlock situations. I had also tried & back tested the 'forced closing' feature which obviously doesn't work as it eats too much into the profits. The moving average also is not effective.

At this juncture, I personally had come to the conclusion that there is no real solution to the deadlock situation and the only way to resolve it is to limit the amount of opened orders. I believe if the limiting orders feature is implemented properly, it should resolve the deadlock situation. I will share my idea in my next post but before I do that, I'd like to say a few things first.

I personally feel that the EA as it is now is unnecessarily overly complicated.

Swingman, please read my idea below. Swingman, can it be implemented into the existing EA ??? I sincerely hope that it can be done. If it is NOT possible or way too complicated or too much work, is it possible to recode it from scratch??? If it is too much work, please do let me know as I do not wish to impose too much of my ideas on you to code them.

Honestly, if for whatever reasons that my idea below cannot be implemented, I personally would not like to pursue this thread any further as I don't see any prospects in resolving the deadlock situations. What is more, the real intention of starting this thread had already been fulfilled - we already have a perfectly working EA (Tic-tac-nur's EA, versions 2.2 & 2.4) trading in the Non-Hybrid mode - which I can use in my live trading as a tool to assist me in my live trading.. But you guys can continue to develop the hybrid mode here in this thread and I will continue to monitor the progress but I will not participate.

Philip Lee May 12, 2018 9:24am | Post# 723

The rationale behind my following idea is to exploit the strengths of both the Non-Hybrid & Hybrid mode.

By implementing it that way, we would have exploited the strength of the trend-bias approach of the Non-Hybrid mode yet at the same time, we will give plenty of room for the Hybrid Mode (which is a very safe way to trade) to bring in the profit but ultimately we set a limit to the opened orders in order to eliminate the possibility of a deadlock situation taking place.

This is the logic flow of my idea:

1. Ascertain the general trend first - whether it is trending up or down;
2. First order must be the same direction of the general trend - thereafter, normal hybrid mode;
3. Limit the number of both the pro-trend & counter-trend orders (eg allow 20 pro-trend orders and 5 counter-trend orders only, or 40 orders to 10 orders etc);
4. When the maximum allowed orders had all been triggered, the EA will immediately start a FRESH INDEPENDENT CYCLE - but the EA must continue to monitor the TP and SL of the maxed-out cycle and must close that maxed-out cycle independently from the fresh cycle (fresh cycle must also be closed independently from the maxed-out cycle);
5. All the independent cycles (maxed-out and fresh) to be closed when either the TP or SL is hit.

I will share my idea of how the above could be implemented in my next post.

Philip Lee May 12, 2018 9:58am | Post# 724

The following are my ideas of how to implement the above logic flow:

1. Determining the general trend

I don't know whether it is possible to code this or not.

The EA will be attached to a particular time frame - the EA must first check the next two (or more depending on the preset parameter) higher up time frames to see if all of the time frames agree on the general trend - and ONLY WHEN THE GENERAL TREND IS CONSISTENT throughout the 3 (or more) time frames that the EA would commence the cycle in Hybrid Mode by starting the VIRTUE PENDING ORDER as usual (ie not executed order) in the general trend ONLY (ie it MUST NOT start a virtual pending order in the counter-trend). Let me give a working example to make it crystal clear.

Let's say we attached the EA to the 4H time frame and we had preset the parameter to check the next two higher up time frames. So, the EA must first check the general trend direction of ALL the 3 time frames of 4H, Daily & Weekly and ALL the 3 time frames must be consistent/agree with each other before the EA commence/issue the Virtue Pending Order. For example, all the 4H, Daily & Weekly time frames must be UP and or ALL must be Down before the EA commence trading.

If all the time frames are UP - then the EA will immediately issue a BUY Virtue Pending Order only. It must not issue a Sell Virtue Pending Order YET, please see below.

If all the time frames are DOWN - then the EA will immediately issue a SELL Virtue Pending Order only. It must not issue a Buy Virtue Pending Order YET, please see below.

There must be an adjustable parameter to determine when the EA could start to issue the counter-trend Virtue Pending Order. For example, lets say all the 3 time frames are UP and we have set the parameter to 3, then the EA must first allow 3 Buy Orders to have been executed/triggered before it is allowed to issue the very first Sell Virtual pending order. I hope this is clear enough.

Thereafter, the maximum number of both the pro-trend and counter-trend orders must be independently adjustable. Continuing with the example above, let's say the general trend as decided by the 3 time frames was UP and if we had preset the respective pro-trend and counter-trend maximum orders to say 40 and 10 respectively, then the EA can only issue a total maximum of 40 Buy orders and a total maximum of 10 Sell orders only.

I hope the above is clear enough.

Philip Lee May 12, 2018 10:08am | Post# 725

As and when the maximum allowed orders have been triggered/executed (both legs) - a total of 50 orders (40 buy and 10 sell) in our previous example - the EA must immediately start a fresh cycle by checking the next two higher time frames again and will only start the fresh cycle when all the 3 time frames agree/consistent with one another - if the time frames do not agree with one another, the EA must keep waiting until all 3 are consistent before the EA starts a fresh cycle. But the EA must continue to monitor the just maxed-out cycle and close it as and when the TP or SL is hit.

Philip Lee May 12, 2018 10:33am | Post# 726

I am sure most of you will agree with me if the above could be implemented, we would have exploited the strength of BOTH the Non-Hybrid and Hybrid modes and would have definitely eliminated the dead lock situation.

The only disadvantage is that some cycles would hit the SL and we must accept losses once in a while. But I am sure most of you would agree with me that if we had started the cycle with all 3 time frames agreeing on the general trend, it would most probably than not that the cycle would end-up/close in profit.

The advantages are as follows:

1. As we are trading with the trend, we can set a much much higher profit target than we could in a standard hybrid mode - for 1.0 full lot size, we could easily set TP of $300 to $1,000 as opposed to a typical $50 TP in a standard Hybrid Mode;

2. With the orders biased/skewed towards the general trend, the cycle would most probably closed in profit way before the max number is reached.

3. We would have traded within the safety of Hybrid Mode most of the time;

4. No precious time would be wasted as the EA would immediately starts a fresh cycle when the max orders have been reached - as opposed to the existing EA which would be stucked in a deadlock situation for 3 weeks just to squeeze out $50 profit;

5. VERY VERY low draw down and in view of the fact that the EA would not trade that often while it waits for all the time frames to agree with one another, we could definitely trade on multiple pairs concurrently;

6. we can trade on any pairs at all as volitility would not be needed since we are trading with the trend;

7. We do not need to do much analysis in Non-Hybrid mode as the 3 time frames needed to agree with one another is already a very powerful filter in itself - in other words, if the STOCH & one-cycle mode can also be implemented, we can trade in Non-Hybrid mode literally 'blindly' ie let the EA do the analysis for us.

There are many other advantages but the above listed are enough to show the advantages of implementing the above. I am sure you all can see & appreciate the above.

SWINGMAN, CAN THE ABOVE BE CODED??? I SINCERELY HOPE SO AND I SINCERELY HOPE YOU ARE WILLING TO GIVE IT A TRY!!!

Philip Lee May 12, 2018 10:47am | Post# 727

The additional adjustable parameters for my idea above would look something like this:

Check Weekly TF - true/false
Check Daily TF - true/false
Check 4H TF - true/false
Check 1H TF - true/false

Initial Pro-Trend Orders to be opened first - 3

Max No. of Pro-Trend Orders - 40
Max No. of Counter-Trend Orders - 10

Start Fresh Cycle after Maxed-out - true/false

SwingMan May 12, 2018 12:47pm | Post# 728

...Honestly, if for whatever reasons that my idea below cannot be implemented, I personally would not like to pursue this thread any further as I don't see any prospects in resolving the deadlock situations. What is more, the real intention of starting this thread had already been fulfilled - we already have a perfectly working EA (Tic-tac-nur's EA, versions 2.2 & 2.4) trading in the Non-Hybrid mode - which I can use in my live trading as a tool to assist me in my live trading.. But you guys can continue to develop the hybrid mode here in this thread...
Philip, thank you for your support so far.

The thread was started to get "a little help" and correct existing EAs. I am glad that with the Tic-Tac-Nur versions until 2.4 the goal was successfully fulfilled.
The other suggestions are indeed very important to implement, but this is a new project, almost "from scratch" where I would have to think and program step by step.

Unfortunately, this does not work for me from one day to the next as in the last time, so I will stop this way of working.
If I'm going to have free programming time in the future, I'll try to take the new suggestions and post them here.

nihilist23uk May 12, 2018 2:01pm | Post# 729

Hi Swingman.
I for one am very grateful of your time with the programming and have to disagree with Philip on the matter that the ea has become to complicated.
It has been on a journey through the versions but the journey has given rise to some great ideas and has helped the ea evolve into a potential gold mine in my eyes.
The features that have not worked can be removed,but we would not have found out if they were no good unless they were programmed for testing in the 1st place.
Thank you.

Version 5.1 is a great ea and just needs a few refinements and I do hope you do not abandon it completely as I feel it has serious potential.
5.1 can delivery 34k return on a 10k account over 11 months,so although the draw down is high by reducing the lot size we can get to double an account with an acceptable exposure level.
I havent even started to optimize yet as my computer power is limited.

Maybe a new thread ?
Look forward to hearing back from you.

SwingMan May 12, 2018 2:44pm | Post# 730

...Version 5.1 is a great ea and just needs a few refinements ...
I am a very old man, and can not always follow Philip's ingenious ideas ...
The last changes (MovingAverage, SwitchMode, StrongPendings) were tryings to reduce the DrawDown. I have only tested for a short time and know that they work in some combinations and not in others.

But I do not invest time to program something like this if Phlip has found that they are basically ineffective and that he does not accept and pursue something like that.
He is 100% right with his ideas and I hope that other programmers will be interested to implement his ideas and to master the DrawDown.
(I have no interest in a new/different thread to make any further developments.)

As for the version 5.1, I try in the next time to find out why there is a difference to 6.1. If I find that, I will post a new version 6.x.

albert666 May 12, 2018 2:44pm | Post# 731

Hi Swingman. I for one am very grateful of your time with the programming and have to disagree with Philip on the matter that the ea has become to complicated. It has been on a journey through the versions but the journey has given rise to some great ideas and has helped the ea evolve into a potential gold mine in my eyes. The features that have not worked can be removed,but we would not have found out if they were no good unless they were programmed for testing in the 1st place. Thank you. Version 5.1 is a great ea and just...
Hello Nihilist,

Can you please share your setfile, i 'm certain it will help a lot of us, sure it will help me

Thanks !

nihilist23uk May 12, 2018 2:53pm | Post# 732

{quote} Hello Nihilist, Can you please share your setfile, i 'm certain it will help a lot of us, sure it will help me Thanks !
Its your set file just look back at post and you can see I just changed trading times.

Innovator757 May 12, 2018 2:56pm | Post# 733

I have seen a video with expert advisors where you just insert the programming blocks. What are the advantages and disadvantage's of expert advisors?

Philip Lee May 13, 2018 2:42am | Post# 734

Hi Swingman,

First of all, I am sure I represent everyone here by saying that all of us truly appreciate what you have given here in this thread so generously and I would like to say a big Thank You to you. Swingman, you are not being paid to do all these, that is why I did not feel it is right to put too much pressure on you to code the additional ideas into the EA. You have already done what you could and nobody can expect anything more from you, just as none of us can expect anything more from Tic-tac-nur. This is the basis of this forum - it is operated on a voluntary sharing basis with no obligation on anyone's part carry the burden all the way to the end. With that, all of us should say a big THANK YOU to Swingman for his selfless contribution here.

So, Swingman, feel absolutely free and 100% justified in wanting to stop at this point - you have already done more than anyone of us in this thread.

Honestly, I was very reluctant to share my idea above as I know it would put tremendous burden and pressure on you - that is why I was frank to ask you to tell me frankly if it is too much to code. I fully understand your position. What is more, as you have stated, this is supposedly a part-time hobby for you, somehting which is supposedly enjoyable and satisfying. I guess with all the pressure on you now, it is no longer an enjoyable hobby to do. That again, I fully understand and Swingman, you will always have my highest respect & regard.

Cheers Swingman, but if in your free time you have managed to code my idea above into an EA, please do share the EA here for the benefit of all.

Thank you so much Swingman for every thing that you have done in this thread.

Philip Lee May 13, 2018 2:45am | Post# 735

Swingman, please feel 100% free to continue to participate in any way you would like to in this thread (if this thread is to be continued).

Cheers Swingman!!!

Philip Lee May 13, 2018 3:09am | Post# 736

Hi Nihilist23uk,

First of all, I must agree with you that this EA (from Tic-tac-nur's V2.2 onwards to Swingman's V6.1) has HUGE POTENTIAL.

This cannot be denied due to the simple fact that the underlining strategy is VERY SOUND.

When I first started this thread, I know that I may not even get a proper working EA at the end of the day as we are 100% reliant on coders to deliver the EA for us, having absolutely zero ability to code myself. That is the risk that we have to take by sharing our ideas here. I do not expect much in return as I know the risk involved. The original intention of starting this thread has already been fulfilled. So, personally I am happy as things stands now. I truly cannot ask for anything more. Both Tic-tac-nur and Swingman had delivered more than we could ask for. I guess if there is no other takers from here onwards, I guess I have to end the thread here.

Are there any takers out there to take this further???

If not, then I guess its the end of this thread and all of us who have truly grasp and understood the strategy behind this EA could take it further privately with paid coders. It cannot be denied, this strategy has HUGE potential but if nobody else wants to take it up, there is nothing further that we can do!!! I certainly cannot code anything. But then again, if I could code the EA myself, I would certainly have NOT shared my idea here. So, I guess everyone here benefited from my inability to code as I have shared my idea here.

I guess I will leave this thread open for a few more days and if there is no further interest, then I will close it.

All of you who had grasped and understood the strategy behind this EA, you have NOT lost anything, you have in fact GAINED a lot. You can now take it further with professional paid coder(s).

Cheers everyone!!!

nihilist23uk May 13, 2018 5:14am | Post# 737

Hi Nihilist23uk, First of all, I must agree with you that this EA (from Tic-tac-nur's V2.2 onwards to Swingman's V6.1) has HUGE POTENTIAL. This cannot be denied due to the simple fact that the underlining strategy is VERY SOUND. When I first started this thread, I know that I may not even get a proper working EA at the end of the day as we are 100% reliant on coders to deliver the EA for us, having absolutely zero ability to code myself. That is the risk that we have to take by sharing our ideas here. I do not expect much in return as I know the...
Thank you Philip for sharing your idea. I too wish I could code !
I have already thanked Swingman and even if it stops here I feel I have learnt something valuable.( Even if its just a huge respect for coders !)

Money is no good without time to spend it and time and health is the most valuable thing we have. Share it but do not waste it.

Thank you to everyone thats help the thread along.

SwingMan May 13, 2018 6:23am | Post# 738

... Cheers Swingman, but if in your free time you have managed to code my idea above into an EA, please do share the EA here for the benefit of all...
Sure, but not if it gets too complicated. Unfortunately for a programmer I have two characteristics: understand slowly but forget fast ...

nihilist23uk May 13, 2018 7:55am | Post# 739

{quote} Sure, but not if it gets too complicated. Unfortunately for a programmer I have two characteristics: understand slowly and forget fast ...

Philip Lee May 13, 2018 9:06am | Post# 740

{quote} Sure, but not if it gets too complicated. Unfortunately for a programmer I have two characteristics: understand slowly but forget fast ...
Ha ha Swingman, that is funny.

I am considering starting a new thread with my newly suggested idea above as the starting point - maybe you can also participate in the new thread as and when and if I decided to start it.

Cheers Swingman!!!


© Forex Factory