NormalizeDouble(Bid, Digits) - page 3

[Alain Verleyen]

whroeder1:
Congratulations, you found a bug in MathRound.

Actually there is no bug in MathRound, you misunderstood.

printf("%.15f %.15f %.15f",price,price/tickSize,MathRound(price/tickSize));

2018.04.05 00:47:16.628    Test EURUSD,H1: 1.700605000000000 170060.499999999970896 170060.000000000000000

Actually it's not a bug at all in my opinion.

[nicholish en]

So can we all unanimously agree that we should no longer say the use of ND is "always" wrong?

[Lakshan Perera]

nicholishen:
So can we all unanimously agree that we should no longer say the use of ND is "always" wrong?

Yes, I do

[Thushara Dissanayake]

Thank you for all. After read all 3 pages i changed my code to this way. Please tell me is this correct or not. Thank you.

double TP_Level = NormalizeDouble(Bid-(pips*Take_Profit),Digits);
double SL_Level = NormalizeDouble(Bid+(pips*Stop_Loss),Digits);
              
if(CheckStopLoss_Takeprofit(OP_SELL,SL_Level,TP_Level) )
{              
  ticket=OrderSend(Symbol(),OP_SELL,LotsOptimizedsell(),Bid,3,0,TP_Level,"macd sample",MagicNumber,0,Red);
}

Earlier in this OrderSend(), I used order open price as "Bid". But when i send it for review, the product is failed by "Error 130". After that i used "ND(Bid)". I only added that ND and it is Ok! So is their any relation?

double ND(double val)
{
   return(NormalizeDouble(val, Digits));
}

 

So finally what you recommend for me?

[Petr Nosek]

nicholishen:
So can we all unanimously agree that we should no longer say the use of ND is "always" wrong?

I've never said that the use of ND is always wrong and I won't start with it. But I'm afraid someone will post this here every day.

[Petr Nosek]

Thushara Dissanayake:

Thank you for all. After read all 3 pages i changed my code to this way. Please tell me is this correct or not. Thank you.

Earlier in this OrderSend(), I used order open price as "Bid". But when i send it for review, the product is failed by "Error 130". After that i used "ND(Bid)". I only added that ND and it is Ok! So is their any relation?

So finally what you recommend for me?

You should adjust price, SL and TP to Tick Size. BTW Error 130 means invalid stops (SL, TP). I use below code to normalize price, SL and TP.

double NormalizePrice(const double price,string symbol=NULL)
  {
   if(price<0.0) return(0.0);
   if(symbol==NULL || symbol=="") symbol=_Symbol;
   double tickSize=SymbolInfoDouble(symbol,SYMBOL_TRADE_TICK_SIZE);
   return(round(price/tickSize)*tickSize);
  }

[William Roeder]

Alain Verleyen: Actually there is no bug in MathRound, you misunderstood.

2018.04.05 00:47:16.628    Test EURUSD,H1: 1.700605000000000 170060.499999999970896 170060.000000000000000

Actually it's not a bug at all in my opinion.

Then NormalizeDouble(170060.499999999970896,0) is wrong and MathRound is correct.

[Alain Verleyen]

whroeder1:

Then NormalizeDouble(170060.499999999970896,0) is wrong and MathRound is correct.

I would say no if we refer to the documentation. For MathRound() it's clearly specified "rounded off to the nearest integer". While NormalizeDouble() documentation is just saying "Rounding floating point number to a specified accuracy", without specifying what rounding.

But that's interesting, maybe you will finish to convince me about NormalizeDouble

Seriously, if a function is working as documented and it's behaviour is coherent, there is no problem to use it. Up to now I never read an example showing an unpredictable or incoherent value returned by NormalizeDouble.

[nicholish en]

whroeder1:

Then NormalizeDouble(170060.499999999970896,0) is wrong and MathRound is correct.

http://www.dummies.com/education/math/basic-math/how-to-round-numbers-up-and-down/

[Thushara Dissanayake]

Thank you for every one!