Major problem with metatrader | Page 2 | Forex Factory

Options

Go to
Permalinks
Print Thread

Similar Threads

Metatrader 4 vs Metatrader 5! 23 replies

Metatrader problem...help!!! 6 replies

Metatrader 5 is going to wipe out your Metatrader 4 indicators / systems 13 replies

Strategy Builder FX Metatrader Problem?? 3 replies

usd/cad problem on MetaTrader 3 replies

Attachments: Major problem with metatrader
Exit Attachments

Major problem with metatrader

  • #21
  • Nov 25, 2009 8:56am Nov 25, 2009 8:56am
It appears that MA's are capped out at 4000 ( ver 225 ) regardless of max bars fwiw.
  • #22
  • Dec 30, 2009 9:42pm Dec 30, 2009 9:42pm
I'm still having this problem, which seems to also exist in any indicator built from MAs(virtually of them). This time I'm using an OsMA indicator and even periods as low as 100 are giving me values that don't match up with the indicator on the chart. And I know I have the history, 1000 minute bars is less than 17 hours of data.

Inserted Code
double zz;
int start()
  {
  int e;
  double Xindi;

  if(zz!=iTime(NULL,PERIOD_M1,1)) {
     Xindi=iOsMA(NULL,PERIOD_M1,12,100,100,PRICE_CLOSE,1); 
     if((TimeToStr(TimeCurrent(),TIME_DATE)=="2009.12.01") && (TimeToStr(TimeCurrent(),TIME_MINUTES)=="01:00")) { Print("Here: "+Xindi); } 
     zz=iTime(NULL,PERIOD_M1,1);  }
  }
  • #23
  • Jan 1, 2010 9:44am Jan 1, 2010 9:44am
Aiccia

Good luck in your quest,I cannot code in this platform at all otherwise I would be helping..

From Horace ...

dimidium facti qui coepit habet

P
  • #24
  • Last Post: Jan 15, 2010 8:37pm Jan 15, 2010 8:37pm
I've no time to test this but have some thoughts on the topic:

I don't know if you found a bug or not, but if it's a bug, most likely it's restricted to backtesting calculations and not real-time charts. However, until it's resolved, there's no way to be sure.


For real-time calculations, what I've observed is:
Suppose you have a 100 SMA (Close_price) on an H1 chart plotted.
Now on an M5 chart, plot a 1200.

FYI, the reason for the "1200" is: 60min/5min = 12, and 12 * 100 = 1200

Compare the SMAs on the two charts. The two SMA's are fairly close, within a small number of pips. Large bars may legitimately cause the values to differ but normally not by much. The fact that they are close leads me to judge that there is no fundamental bug with large-value MA's, at least for real-time plots. (Maybe I haven't looked close enough like going way far back in time on a chart to compare; it's been awhile since I last checked this).


Another thought...
Consider how an SMA might be calculated. There are two basic approaches, but one is more likely because it's less computation (but possibly more prone to error over long periods).

To calculate a 100 SMA, one could literally add up the previous 100 bars price, divide by 100 and get the value. Then on to the next bar, do exactly the same thing, ... This method works but is highly inefficient.

An alternative efficient method would be to calculate the first 100-SMA value on the chart (might be the most recent bar, might be the oldest; just start somewhere. Below assumes Bar(0) start). Since you know that bar's 100 SMA value, what is the SMA value for an adjacent bar?

First bar calcuation:
sma(0) = (add the last 100 bars price)/ 100 ;

Then to get all the other 100 SMA values on the chart:
for (n = 1; n < Bars-100; n++)
{sma(n) = ( sma(n-1) * 100 + Close(n+100) - Close (n-1) ) / 100; }

In other words, take the first calculated SMA value, multiply it by the period (100) to get the total sum of prices, subtract the previous n-1 price but add the n+100th price to get a new SMA total-sum, then divide again by 100 to get the sma(n) value).

This later method is almost certainly the method used because it is computationally faster. However, it may well be subject to round-off inaccuracies over a large number of bars, esp. for large SMA values.

Just some thoughts... hope you resolve the problem, and if it truly is an MT bug, it should be reported to them and fixed.
0 traders viewing now