Help Needed to Debug Existing EA with proven strategy

Hi,

Way back in 2011 when I actively traded in forex I had spent quite a bit of $$ employing coders from Scriptlance to code my ideas into EA. I had stopped trading in 2013 due to huge losses but am thinking of trading live again in the near future. I recently re-looked into the arsenal of EAs that were coded for me, I discovered some gems and would like to invite all coders here to help me debug one of these gems for the benefit of all - I will post the source codes here. The EA is called Forex RM split into a Buyer and a Seller. RM stands for Reverse Martingale. Don't worry, it is NOT a Martingale nor is it a pure reverse Martingale - I would describe it as a hybrid reverse Martingale. Honestly, the simplicity of the strategy will make you laugh but believe me, IT WORKS VERY WELL in practical reality - I had tested it extensively both in demo and live accounts. The only reason I had stopped using it on my live account is the bug that really cripples it and I was not focused nor determined enough to iron out the bug at that time as my head was full of ideas and I was impatient to get them all coded.

In order to stir up your interest, let me state the following categorically:

1. It is VERY PROFITABLE
2. It is very low risk
3. Hardly any draw down
4. It will never blow your account (so long as SL is set)
5. It can be traded on all the 23 pairs simultaneously
6. If employed properly, it can cope with ANY market situations - trending, ranging except perhaps crazy market situation
7. If the bug is successfully fixed and the EA could trade in the manner that I wanted/intended it to, there is absolutely no reason why you can't make at least 3% per trading day trading on all the 23 pairs;

I will post screen shots here to show that in the past 44 hours trading from 28.03 at 01:40 to 29.03 at 21:30, it made $12,469.00 profit (out of a $100,000 demo account) representing 12.47% in less than 2 full days trading. It had actually made $16,652.59 (16.65%) but I had manually closed -$4,183.59 worth of trades due to the bug - trades which should have been closed in profit but was not closed because of the bug. It can also be seen from one of the screen shots that out of the -$4,967.41 drawdown from the on-going active trades, -$4,841.33 of which were caused by the bug - the other words, the actual drawdown is only -$126.08!!! (and that is after having actually made $16,652.59 in those 44 hours). Don't worry about the huge $100,000 of the demo account - it was meant to simulate a $1,000 micro/cent account since we can't get demo account on micro/cent account ($100,000 in a standard account equals to $1,000 in a micro/cent account). This is a screen shot of a typically & properly closed trade cycle when Universal TP set at $300.

Here are the EAs and the original description file which I gave to the coders to explain my idea to them.

These are my rules in this thread:

1. I would like to iron out the bug from these two EAs first and make them perfect first before we move on to combine the two into a single EA with a trend filter coded in so that it can be traded fully automated (the split Buyer & Seller currently needs some manual input which will be explained in my next post)

2. so, please refrain from anything else but stay focus on debugging the Buyer and Seller EA first

3. Later once they are perfect then we can discuss on what filter to incorporate in order to combine them together and to get it to operate fully automated

4. Only well-intention coders please

5. Please refrain from disrespectful remarks

These are the universal settings used on both the Buyer and Seller on all the 23 pairs which were all traded simultaneously

Let me explain the simple strategy behind this simple EA first before I go on to describe how I trade it.

The EA differs from a typical Reverse Martingale strategy in that the First Order is designed to seek out the trend and or seek to get to as close as possible to the bottom/top of a ranging market before kick starting the Reverse Martingale strategy. We will always use the smallest size of 0.01 lot for the First Order. TP of the First Order will always be set at 0 - this TP setting feature can actually be eliminated as it is redundant as we will always set the TP to 0. The First Order can best be described as "expendable trend or bottom/top seekers" - and in practical reality, we will get lots and lots of small losses when the First Order hits its SL - we in fact want the First Order to keep hitting the SL until the trend reverses and or when the bottom/top had been reached and when it does so, the Reverse Martingale will kick in and the big profits will roll in pretty quickly. As can be seen from one of the screen shots I posted above that it typically takes only two subsequent order to be triggered in order to hit the Universal Profit set at $300.

Of course we will get a situation whereby subsequent order(s) was/were triggered but the trend reverses before the Universal TP is hit and eventually hit the SL of the First Order (this is the only situation where draw down will be incurred) - and this is where the bug is - when this happens, the EA must immediately re-issue a fresh First Order at the then market price together with the two fresh stops pending orders. This is where the bug is - that the EA always failed to re-issue the fresh First Order & the two pending orders. As the triggered/opened subsequent orders are 100% reliant on the First Order and the following subsequent triggered orders to recover, they will be left hanging as 'orphaned' trades and eventually turn into toxic trades.

I will discuss on the bug in the coming posts.

Hi reteid2222,

A coder by the name of Irwin (from Indonesia) coded it for me in 2011 - please see the copy right and his email address in the source code.

Hi reteid2222,

Can we do it through this thread instead of through skype as I would like everyone to benefit from this. Thanks very much for the interest anyway.

I will describe the simple approach I take to trade the EAs in my next post.

I will look at the D1 candles shortly before or after it closes and with the help of Bollinger Bands & STOCH (or any indicators of your choice) - I will just roughly guess what direction the price will move in the next 24 hours - if I think it will go up, I will employ the Buyer EA & vice versa. NEVER employ both Buyer & Seller on two separate charts at the same time - its idiotic to do so. Don't worry about getting the direction spot on - the EA is very flexible and can cope with practically any situations. Once that is done, I will just let it run for the next 24 hours. The next day, I will do the rough guess again and in most cases, no changes will be needed - but if I think the trend had reversed, I will close up all the open trades on that pair (you can well afford to do that due to the very nature/strategy employed by the EA, the losses incurred by force closing them are well covered by the huge profits already made by then - difficult for me to explain, try it and you will know what I mean).

I will try to clean up everything on Friday but if the drawdowns are substantial, I would just leave them as they are - the EA can easily cope with them when reopen again on Monday.

Its that simple really.

Try it out and you will know what I mean.

This is the bug which I would like to invite the coders here to iron out.

The bug only occurs in a situation when the Subsequent Order(s) was/were triggered - if the Subsequent Order(s) was NOT triggered, the EA functions properly.

This is the scenario when the Bug surfaces 100% of the time - it is when the Subsequent Order(s) was/were triggered and before the Universal TP was hit, the trend reverses and eventually the First Order is closed after having hit the SL - the EA always failed to reissue a fresh First Order (together with the two stops pending orders) when this happens.

If you had grasp and truly understand the strategy employed here, you would immediately understand how VITALLY IMPORTANT that the EA must immediately re-issue a fresh First Order (together with the two fresh pending orders - and of course, it must delete all the untriggered stops pending orders first).

This is where I need your help - could you please debug it and make sure that the EA ALWAYS & IMMEDIATELY issue a fresh First Order eveytime that the previous First Order has been closed by hitting the SL.

The First Order MUST always be re-issued, otherwise the whole strategy is defeated.

I hope you guys can help me out here and by so doing, everyone also benefit from this - believe me, this EA is VERY PROFITABLE and low-risk & it will never blow up your account - so long as you always set the SL for the Subsequent Orders.

Hope to hear from you guys.

Thanks very much.

The concept of the first order as a factual position (executed order) is redundant. Instead it can exist purely virtual within the algorithm, not causing actual loss when its virtual stoploss is 'hit'.

This

I agree with rockit and Nicholishen.

Also, given what you say in post #8, if you can exactly define your strategy (D1 candle, stoch, BB, whatever) that chooses either the Buyer or Seller EA, then all of the Buyer, Seller and 'master selection' processes could be combined into a single EA, both tidying up and automating the entire process.

I've just read the docx and I've seen strategies using similar principles to those described. That's all I'll say.

Hi rockit, Nicholishen & Hanover,

Thanks very much for your reply. Wow I must admit that I am not tech savvy and had absolutely no knowledge whatsoever in coding - and please understand that I had just came back to the Forex scene about one month ago and am trying to relearn everything again.

Wow guys, it is literally shocking to me to learn the fact that the First Order can be done virtually within the algorithm - honestly, none of the coders that I engaged extensively with (more than 10) in Scriptlance back in 2008 to 2012 had ever told me that it can be done virtually - I had at least 20 EAs coded by them and most of which are riddled with bugs which in most cases, despite clear explanation of the bugs, they were not ironed out properly (such as the present EA which I posted here).

As stated in post 3, I would like to do the combined EA eventually as suggested by you guys but again, as stated in the same post 3, can you guys help me iron out the bug first in the existing Buyer and Seller EAs???

Maybe I belong to the old school thinking (am 53 years old this year) - I'd like to see with my naked eye the EA execute the strategy as I intended it and leaves behind a track for me to trace & follow up in order to find where the bugs are and also to verify whether or not the strategy actually works. If everything is done virtually, I'd think its a bit more difficult (to me at least) to trace and to verify the validity of the strategy. Another reason for having requested the original coder to code two separate Buyer and Seller was because I want to have a certain degree of control over the EA in particular the entry.

So, can I request you guys to be patient and bear with an old school thinker and help me iron out the bug which I had described in post 9 above?

Don't get me wrong, I am VERY interested in the combined EA where most features are implemented virtually within the algorithm but can we please make the two Buyer & Seller EA perfect first???

Once again, thanks very much guys for the interest shown.

Hi hanover,

Its an honour that you replied to my post. I had read numerous of your posts way back in 2008 - 2012 when I was actively lurking here (but did not actively participate) and I can vividly remember your username 'hanover' as it is easily rememberable. I also had two or three short exchanges with nanningbob then, another easily remembered username when he was still then heavily into Martingale strategy and was attempting to find ways to make it safer - I noticed and read quite a bit at Stevehopwood site recently that he had moved forward tremendously - even back then, I could clearly discern that he is one hell of an altruist guy.

Re your suggestion in your post (12), I am all for it but as stated in post 3, could you please help me iron out the bug first and just perfect the EA as it is???

Hanover, as you stated , I agree that I could not be the only person who had thought about using the First Order as a trend/bottom/top seeker, but I can state this with 100% honesty, I did not read it nor copied it from somewhere - it was 100% my original idea - but I have no problem accepting the fact that others must have thought about it as well. I had read many other 'hybrid' ideas in as far as reverse martingale is concerned but I have not read one like mine (I am sure someone somewhere had written about it but I definitely had not come across it). Never mind whether its my original idea or not, I am not egoistic at all, the most important thing is that it works and if you guys can help me make them perfect, numerous others would benefit from it.

Thanks again hanover for the interest shown!

Hi hanover,

In further reply to your post (12), allow me to explain further why I would dearly like to have a perfected separate Buyer and Seller EA first before moving forward with a combined EA.

Firstly, please do understand that I have zero knowledge in coding. If I were a coder myself, I would immediately agree and do like you suggested (because I could keep modifying it as my idea evolves/improves).

The issue here to me as a non-coder is the entry - if the entry conditions are hard coded in the EA, there is nothing that I could do (as I do not know how to modify it) if it turns out to be ineffective. Please do understand the fact that our knowledge on forex evolves and keep improving - let me take my own example here - as stated earlier in post 8 that I currently use the Bollinger Bands, STOCH and candle patterns to help me roughly guess where the direction would be heading in the next 24 hours - but just a few days ago I came across Bernhard excellent Advanced Dashboard for Currency Strength (please go to the commercial section) which have many excellent reviews - and as and when the EAs here are perfected by you guys here, I will definitely use his indicator to see the strength and weakness of the respective currency in a particular pair in order to decide whether to employ the Buyer or the Seller for that particular pair.

In short hanover, I just want the EA to execute the trade once I had made the filtering decision - if the entry condition is hard coded in the EA, I would not have that freedom to improve on the entry - I am sure 90% of those who are reading this who are not coders themselves appreciates what I have written above.

Believe me, if the filtering process is done properly (Bernhard's excellent indicator would help tremendously in this respect), the Strategy WORKS VERY WELL in reality - of the months of testing the EA (back in 2011) both in demo and live account, I have had not a single Subsequent Order hitting the SL (set at 150 to 200).

I dearly sincerely hope that you guys can help make these two EAs perfect first as many would benefit from it.

Hi Phillip

Trading the EA on a demo account is likely to hide the dangers behind this technique.
How does the EA perform in a live market and during news spikes?

Let's have a look at a possible scenario where the market is moving against us.

At -10 pips the 1st suborder triggers so we have 1.01 lots open
At-15 pips the SL of the fist order gets hit so we have 1.0 lots open

In a live situation where the SL of the 1st order gets hit,
will the EA have time to delete the 2nd pending order before it triggers and becomes a liability?

If it doesn't and our 2nd suborder is only 5 pips away, it could also trigger and then we will have 2 lots open
and our suborder SL is approx 200 pips away potentially exposing us to a total loss of 410 pips plus slippage so perhaps a -$4500 loss
If the EA does have time to close the pending order then we only risk 200 pips +slippage so perhaps our loss due to a news spike will be -$2500

Hi Billy,

As stated in my earlier posts, the filtering process is the key to this strategy - that is precisely why I want the two EAs to be separated and only serve to execute the trades once we had made that filtering decision.

I am sure those of you who are more discerning here would have discerned from my posts that I fully intended to use it only as a tool to assist in my trading - due to its highly flexible approach, I could rely on it to make the trade profitable for me even if my filtering process and my entry was not spot on.

I had ran the demo account and posted the screen shots here trading on all 23 pairs 'blindly' just to show you guys how flexible the approach is and how well it can assist you to turn a not so well entered trade into a profitable trade. The purpose is just to show you guys how profitable and flexible that approach is, nothing more - in reality I would be very cautious to trade all 23 pairs blindly like that unless you have the free time to study each pairs in depth first. I hope you guys understand what I am trying to say here.

That is why I insisted from my Thread rules that the two separate EAs must be made perfect first. I know people will immediately suggest that the two be combined and a filter be hard coded in - from my experience of having testing thoroughly almost 100 EAs, I can tell you for a fact that it will never work because it takes out the most important element of a human filtering process - it is simply impossible to code in all the filters and the constantly changing variables which only a human person can make.

So Billy, the simple answer to your question is not to trade when there is an upcoming 'red' event (check it here at ForexFactory). But as can be observed, during those 44 hours testing, there were two 'red' events but I could not be bothered because I know the EA can cope with it. The other option is to set your SL lower say, 50 or 100 - again, as repeatedly stated, if the filtering is done properly, the SL would be rarely hit, though it would get hit once in a long while.

Billy,

After rereading your post, I would like to further state as follows:

In as far as deleting pending orders are concerned, it should not happen if the EA is functioning properly (but the bug here does failed to delete the same and to reissue the fresh First Order). If the market is crazy and the EA could not cope, well I guess you just have to delete them manually. You obviously would not leave it unattended if you know that news events are forthcoming. The scenario you had described is highly unlikely to happen in a normal situation except in crazy news event (which you should not have traded in the first place) because it assumes a drastic reversal where the price had gone in BOTH directions in a substantial manner - which in reality would have close the cycle in profit - bearing in mind that the pending orders are stops order ie they only get triggered when the price moves in that direction. Billy, very difficult for me to explain here, try it and you will know what I mean - the scenario you had described ie the price had ranged substantially in both direction is actually FAVOURABLE to our strategy - all the trade would have been closed in profit.

I'm guessing this was due to market conditions and not necessarily due to a bug in your EA
in which case even pressing X in Metatrader to close the trade might not work either.

This can happen, your live MT4 can slow down or your broker's server freezes and you cannot close a trade
and it will not necessarily be during red news events.

What does the EA do next, when the SL of the 1st order gets hit and the 1st suborder is already triggered?

Hi Philip,

Many thanks for your reply and explanations. I understand your reasons for wanting to keep the two EAs separate, and to initiate them using human intervention.

I'm afraid I can't help you with the coding. I prefer not to modify other programmers code. Instead, I do all my coding using Metaeditor/compiler build 509, and code libraries that I've developed and tested exhaustively myself.

There are many talented programmers who contribute to this forum. If you can demonstrate that your strategy is profitable, I'm sure that you'll have no problem persuading any of them to modify the EAs for you.

I wish you every success.

David