How do you trust backtesting with MT5? - MQL4 and MetaTrader 4

Sergey Golubev 2013.01.14 08:08 #81

how can you not know this and at the same time tell us how wonderful MT5 is ? it does nothing for your credibility.

And you are telling me how wonder MT4 strategy tester is :) :)
Seems you are going to catch me on behalf of some group of people? :)
I am not a fish :)

Many traders do not know this "At every-tick modeling, tester pumps all necessary timeframes for the symbol under test independently",
and many traders do not trust it.

Moderators of some forums are banning the members for selling EAs based on backtesting results only (especially MTF EAs).
Because it was a lot of cases when coder coded EA and sold it on internet, EA had different results in trading compare with backtesting, buyers lost money, coder disappeared.

Did you compare backtesting results of some EAs with forward testing results?
I did it for many years with many brokers (MT4). Elite section of some forum was started many years ago only because backtesting results of EAs were different from forward testing results in some cases (I described those cases in some posts above) Before elite section - metatrader yahoo group was started with same situation.

As to MT5 - I am trading manually only for now. But it may be good to compare backtesting with trading on the same way as I did it for MT4.
For example - by taking some EAs (EA coded on close bar, EA coded using icustom, EA coded on close and open bar, EA coded on open as icustom, EA coded on high/low of the bar as icustom, EA coded in MTF way etc) and compare backtesting with trading for same period with same broker ... I did it for MT4 some years ago.
The problem is that this experiment will go to nowhere: people will continue using backtesting results as a proof that their EAs are profitable before selling them, and nothing will be changed in anyway.

I am using strategy tester for MT4 (and MT5) to find possible settings for EAs.

By the way, I used Cloud to optimize the settings for MT5 EA yesterday and it took 10 minues for me instead of 10 hours (because of Cloud).
Yes, Cloud was wonder.

Strategy Testing - Algorithmic Real and Generated Ticks Strategy Optimization - Algorithmic

Simon Gniadkowski 2013.01.14 08:29 #82

newdigital:

And you are telling me how wonder MT4 strategy tester is :) :)
Seems you are going to catch me on behalf of some group of people? :)
I am not a fish :)

I make my comments on behalf of myself alone, everyone else can speak for themselves. You came here actively promoting MT5, it is up to you therefore to be the expert on what you are promoting and be able to answer the arguments against it. You can't do that without knowing your facts.

When you compared forward tests vs Strategy Testing, how long was your forward testing ? how any currency pairs did you use ?

Market App Store MQL5 Wizard: Development of MetaTrader Market - the

Sergey Golubev 2013.01.14 08:43 #83

For how long?
I did it since 2006 for elite section of some forum having 5 dedicated servers trading more than 100 EAs. Some metatrader account was started in 2006 and was continuing till 2012 (forward testing).
Did you compare backtesting results with forward testing for MT4? or you just talking about article as the fact?

I opened this thread to discuss the article (see my first post) - the article uploaded in mql4.
It is my goal here.
As to title of this thread so it is my idea.
And I am trying to use MT5 in most of the cases - my idea too. No one told me (and no one can tell me): "for now - you will be busy with MT5 instead of MT4".
And it is really strange to see how some coders (my friends) are replying to me for my email or to my facebook page, ... and they do not want to go to this mql4 forum to reply ...

Messages - MetaTrader 5 Messages - MetaTrader 5 Mailbox - MetaTrader 5

Point Zero 2013.01.14 09:18 #84

IMO, the worst handicap of Metaquotes is no compatibility with past code (MQL4 does not work at all on MT5). It is a very very bad business decision.

I have also developed Apple (Mac, Ipad and Iphone apps) and my applications still work if Apple updates their software. It is a serious company.

For some reason, Metaquotes has decided that the work of their customers has little value. And the price is having no users upgrading to mt5.

I don't use MT5 because I want to use my tools, which are already coded and I don't want to code thousands of lines all over again.

But for a moment, suppose I did all the work again and translated all my code to MT5. Soon enough, MT6 would be released trashing my work again.

So, it is like a vote. And I vote MT4 or having MT4 compatibility.

Account Connection - Accounts Installation on Mac OS Price Charts, Technical and

Sergey Golubev 2013.01.14 09:22 #85

Well ... this case may be simple one:

extern string     Main_data      = " Trade Volume & Trade Method";
extern double     Lots           =     0.1;
extern double     TakeProfit     =     1000;    // Take Profit Value 
extern int        StopLossMode   =     1;    // Stop Loss Mode 
extern int        TrailStopMode  =     2;    // Trailing Stop Mode 
extern double     BreakEven      =     50;    // Break-Even Value
extern int        SessionStart   =     0;    // Start Hour of Trade Session 
extern int        SessionEnd     =    23;    // End Hour of Trade Session
extern bool       SignalMail     = false; 

extern string     Inputs         = " BrainTrend parameters ";
extern int        TimeFrame      =    240;    // Large Time Frame in min
extern int        NumBars        =   500;
extern int        ConfirmBars    =     2;
extern double     Use_pSAR       =     1;
extern double     pSAR_Step      =  0.02;
extern double     pSAR_Maximum   =   0.2;
extern double     Use_Stoch      =     0;
extern int        Kperiod        =    14; 
extern int        Dperiod        =     5;
extern int        slowing        =     5;
extern double     UpLevel        =    75;
extern double     DnLevel        =    25;
extern double     Use_iTrend     =     1;

and

for (int i=num;i>=1;i--)
   { 
   double hiBT1 = iCustom(Symbol(),0,"BrainTrend1",NumBars,1,i); 
   double loBT1 = iCustom(Symbol(),0,"BrainTrend1",NumBars,0,i);
   double hi = High[i];
   double lo = Low[i];
   if (mode==1 && result>=0 && (hiBT1 != lo && loBT1 != hi)) {result+=1;}
   //else {result=0;break;}
   if (mode==2 && result<=0 && (hiBT1 != hi && loBT1 != lo)) {result-=1;}  
   //else {result=0;break;}
   }
   
   return(result); 
}

int TradeSignal()
{   
   
   double buyBT1  = iCustom(Symbol(),TimeFrame,"BrainTrend1Sig",NumBars,0,0,1,1);
   double buyBT2  = iCustom(Symbol(),TimeFrame,"BrainTrend2Sig",NumBars,0,1);
   
   double sellBT1 = iCustom(Symbol(),TimeFrame,"BrainTrend1Sig",NumBars,0,0,0,1);
   double sellBT2 = iCustom(Symbol(),TimeFrame,"BrainTrend2Sig",NumBars,1,1);
   
   if (Use_pSAR > 0)
   {
   double SAR = iSAR(NULL,TimeFrame,pSAR_Step,pSAR_Maximum,1);
      if (SAR < Close[1]) int pSAR =  1;
      else
      if (SAR > Close[1]) pSAR = -1;
   }
   
   if (Use_Stoch > 0)
   {
   double StoMain = iStochastic(NULL,TimeFrame,Kperiod,Dperiod,slowing,MODE_SMA,0,MODE_MAIN,1);
   double StoSig  = iStochastic(NULL,TimeFrame,Kperiod,Dperiod,slowing,MODE_SMA,0,MODE_SIGNAL,1);
      if (StoMain > StoSig && StoMain < DnLevel && StoSig < DnLevel ) int Stoch = 1;    
      else
      if (StoMain < StoSig && StoMain > UpLevel && StoSig > UpLevel ) Stoch = -1;
   }

I mean - EA should be attached to H1 chart for trading and taking the value of some indicators on H4 timeframe.
And we see iTrend indicator ... I spent few years for public developing Brainwashing trading system with some good coders ...
This indicator may show different value for different brokers.

Parabolic SAR - Trend Moving Average - Trend Moving Average - Trend

Sergey Golubev 2013.01.14 09:28 #86

And what about this one?
It is the code from Electra EA

if (UseClassicalEntry)
         {
            string M5_Trend  = findTrend("inf.info10",PERIOD_M5 ,MAPeriod1);
            string M15_Trend = findTrend("inf.info8" ,PERIOD_M15,MAPeriod1);
            string H1_Trend  = findTrend("inf.info6" ,PERIOD_H1 ,MAPeriod2);
            string H4_Trend  = findTrend("inf.info4" ,PERIOD_H4 ,MAPeriod3);
            string PH1_Trend = "";
            string PH4_Trend = "";

            double PvBarMA1=iMA(NULL,PERIOD_H1,MAPeriod2,0,MAMode,MAPrice,2);
            double LastOpen1=iOpen(NULL,PERIOD_H1,1);

               if (LastOpen1 > PvBarMA1) PH1_Trend="UP";
               if (LastOpen1 < PvBarMA1) PH1_Trend="DW";
      
            double PvBarMA2=iMA(NULL,PERIOD_H4,MAPeriod3,0,MAMode,MAPrice,3);
            double LastOpen2=iOpen(NULL,PERIOD_H4,1);

               if (LastOpen2 > PvBarMA2) PH4_Trend="UP";
               if (LastOpen2 < PvBarMA2) PH4_Trend="DW";
            
               doBuy  = (M5_Trend=="UP" && M15_Trend=="UP" && H1_Trend=="DW" && PH1_Trend=="UP" && H4_Trend=="DW" && PH4_Trend=="UP");
               doSell = (M5_Trend=="DW" && M15_Trend=="DW" && H1_Trend=="UP" && PH1_Trend=="DW" && H4_Trend=="UP" && PH4_Trend=="DW");
         }
         else
         {
               doBuy  = (iCustom(Symbol(),240,"AbsoluteStrengthForEA",0,1)==1 && iCustom(Symbol(),240,"AbsoluteStrengthForEA",0,2)!=1);
               doSell = (iCustom(Symbol(),240,"AbsoluteStrengthForEA",1,1)==2 && iCustom(Symbol(),240,"AbsoluteStrengthForEA",1,2)!=2);
         }

and this one ... it is related to forward testing, right? It is not about backtesting ... As I remember - this EA should be attached to H1 timeframe ....

if (iBars(Symbol(),PERIOD_M5)<200)
              { showTwoStateMessage("lessbars5",Symbol()+" waiting for 200 bars on M5",false); break;  }
         else   showTwoStateMessage("lessbars5",Symbol()+" Bars OK for M5 timeframe",true);

         if (iBars(Symbol(),PERIOD_M15)<200)
              { showTwoStateMessage("lessbars15",Symbol()+" waiting for 200 bars on M15",false); break; }
         else   showTwoStateMessage("lessbars15",Symbol()+" Bars OK for M15 timeframe",true);

         if (iBars(Symbol(),PERIOD_H1)<200)
              { showTwoStateMessage("lessbars60",Symbol()+" waiting for 200 bars on H1",false); break; }
         else   showTwoStateMessage("lessbars60",Symbol()+" Bars OK for H1 timeframe",true);

         if (iBars(Symbol(),PERIOD_H4)<200)
              { showTwoStateMessage("lessbars240",Symbol()+" waiting for 200 bars on H4",false); break;}
         else   showTwoStateMessage("lessbars240",Symbol()+" Bars OK for H4 timeframe",true);

Sergey Golubev 2013.01.14 09:35 #87

flaab:

IMO, the worst handicap of Metaquotes is no compatibility with past code (MQL4 does not work at all on MT5). It is a very very bad business decision.

I have also developed Apple (Mac, Ipad and Iphone apps) and my applications still work if Apple updates their software. It is a serious company.

For some reason, Metaquotes has decided that the work of their customers has little value. And the price is having no users upgrading to mt5.

I don't use MT5 because I want to use my tools, which are already coded and I don't want to code thousands of lines all over again.

But for a moment, suppose I did all the work again and translated all my code to MT5. Soon enough, MT6 would be released trashing my work again.

So, it is like a vote. And I vote MT4 or having MT4 compatibility.

yes, flaab
you are right.

As to me ... as I am not a coder (and I never wanted to be a coder) so some things are not important for me. For example, I can understand something inside mql4 code (as every trader should understand). But I understand nothing inside mql5 code - I can understand just a logic only. So, the distance from "understand something" to "understand nothing" = nothing :) I was talking about mql4 and mql5 programming language - if I am not a coder so ...
I like MT5 because of many things (I described it on this thread).

But main my concern is the following: I need to convert all my tools to mql5 programming language.
Yes, I started to do it (it takes money) ... but I am sorry - if it is my tool so why should I send them to any coder with a question "please convert it for ... credits"?
I want to keep it for myself ... I do not want to send it to any coder (with money) :)
But to keep it for myself - I will have to learn mql5 ...

Do you see my image on this page https://www.mql5.com/en/forum/142858/page5 (image with white background)?
This is AccountAnalyzer_v1.1 indicator making this stats from account history.
How can I convert it to MT5 if I do not want to disclose this tool?
No way ... I will have to learn mql5 language anyway ...

I was reading some threads or article that someone proposed to have some MT4 emulator on MT5 ... so I have no idea - it is true or not.

Articles on the development MQL5 Wizard: Development of Algorithmic (automated) trading in

Simon Gniadkowski 2013.01.14 10:33 #88

newdigital:

For how long?
I did it since 2006 for elite section of some forum having 5 dedicated servers trading more than 100 EAs.

I was asking how long did you test an EA for on a forward test compared to the same EA tested with the same settings and comparable spread on a Strategy Tester test ? So how long on the ST ? how long on a forward test ?

I have not tested any EAs seriously on a forward test, I don't have anything that is worth the effort yet.

Close By - Trade Close By - Trade Real and Generated Ticks

Simon Gniadkowski 2013.01.14 10:33 #89

newdigital:

And it is really strange to see how some coders (my friends) are replying to me for my email or to my facebook page, ... and they do not want to go to this mql4 forum to reply ...

What does your personal emails have to do with mql4 coding ? are you getting off topic again ?

Simon Gniadkowski 2013.01.14 10:34 #90

newdigital:

Well ... this case may be simple one:

and

I mean - EA should be attached to H1 chart for trading and taking the value of some indicators on H4 timeframe.
And we see iTrend indicator ... I spent few years for public developing Brainwashing trading system with some good coders ...
This indicator may show different value for different brokers.

What is your mql4 question about this code ?

Why some great coders and trading system developers are ignoring Metatrader 5? - page 9