Metatrader known bugs ... - page 10

Bob Fisher
316
Bob Fisher  
mladen:
Metatrader 5 editor is a rather unfortunate creation. Makes me wonder if anybody of the creators of it ever tried it on real coding example (writing it from first to last letter)

Never happened, the designers are in a dream world all of their own, its how they imagine the world should be, and do not comprehend how it really is. Microsoft's Visual Studio is the same way, developed for designers not developers by other designers.

Mladen Rakic
136225
Mladen Rakic  
Tzuman:
Never happened, the designers are in a dream world all of their own, its how they imagine the world should be, and do not comprehend how it really is. Microsoft's Visual Studio is the same way, developed for designers not developers by other designers.

My favorite is case sensitive autocomplete in metatrader 5 - you even have to remember if you wrote it in capital or small letters if you want the autocomplete to "help" you :):)

Mladen Rakic
136225
Mladen Rakic  

Well we shall have what to talk about in the next couple of months (years) ... with the new metatrader 4 ... unfortunately

BigBe
228
BigBe  

One Click Trading Bug

In FXDD build 509.

It is NOT enabled in Tools /Options / Trade, but it is available on every chart.

I don't want to hit One Click Trading by accident so I want it NOT available when not available.

Big Be

Mladen Rakic
136225
Mladen Rakic  

A bug with metatrader 4 built in SMA (simple moving average)

Attached sample code is showing that SMA is not calculating same values for averages if you change the starting point for calculation even for just one bar. The example bellow has a starting bar set to 4999 in the first example and 4998 in the second example and in the second example the SMA never reaches 0 (and it should since there are series of 0 values longer that the calculating period).

So be careful even with the mots common built in functions : after all these years it is still not sure that they work as they should

Files:
sma_test.mq4 2 kb
sma_test.gif 67 kb
BigBe
228
BigBe  

mladen,

I ran into a similar problem. I had an occasional invalid indicator result (used in an EA), that I finally deduced was from a difference in the 7th or 8th decimal point or thereabouts. I had to re-code to check if the difference or value was greater than 0.000001 or some such. Anyway I learned not to trust the furthest decimal values. 0.00000001 is small but not equal to 0.0. I will try to find that example in my files if anyone needs it.

Big Be

Mladen Rakic
136225
Mladen Rakic  
Big Be:
mladen,

I ran into a similar problem. I had an occasional invalid indicator result (used in an EA), that I finally deduced was from a difference in the 7th or 8th decimal point or thereabouts. I had to re-code to check if the difference or value was greater than 0.000001 or some such. Anyway I learned not to trust the furthest decimal values. 0.00000001 is small but not equal to 0.0. I will try to find that example in my files if anyone needs it.

Big Be

Big Be

It is not me that is doing that in that case (I am not using any roundings or normalzation or anything of a sort). What happens is simple : if you start SMA calculation from the last (oldest) bar using built in SMA, it will calculate OK (it will calculate 0 values correctly where they should be). But if you calculate SMA from the penultimate bar it will NEVER calculate a result of 0 even if it must. NEVER EVER. And that is a bug

SMA is simple : if there are n same values and you calculate a sma of n period, at the end of that n stream the SMA value MUST me equal to that repeated value. Not a bit greater or a bit smaller. It must be equal.

_________________

To resume : SMA works OK when it calculates all the bars. But if you leave out just one single bar, just one, it will not calculate OK (in the example from my previous post it will NEVER calculate a correct value, there will ALWAYS be a deviation from a correct value unlike the case when all the bars are taken into calculation)

Mladen Rakic
136225
Mladen Rakic  

This is, for the first time, not a post about some metatrader bug

This is a result of not knowing what was going on with one EA that was using a multi time frame indicator. As it turns out, metatrader is now simulating current prices in a target time frame too. What it means that even in a back test, if called from an EA, indicator is going to "behave" as in run-time (no more "knowing the future" for indicators). Attaching a simple EA that only shows daily close, high and low and whatever time frame lower from daily you call it from, it will show correct values for those prices (not the future values). The ea should be set to visual mode and then it will display those prices

For once, something that is done as it should be done

Files:
test
1232
test  
mladen:
This is, for the first time, not a post about some metatrader bug

This is a result of not knowing what was going on with one EA that was using a multi time frame indicator. As it turns out, metatrader is now simulating current prices in a target time frame too. What it means that even in a back test, if called from an EA, indicator is going to "behave" as in run-time (no more "knowing the future" for indicators). Attaching a simple EA that only shows daily close, high and low and whatever time frame lower from daily you call it from, it will show correct values for those prices (not the future values). The ea should be set to visual mode and then it will display those prices

For once, something that is done as it should be done

Does that mean that multi time frame testing in EA can use an opened bar now?

Mladen Rakic
136225
Mladen Rakic  
checkin:
Does that mean that multi time frame testing in EA can use an opened bar now?

Yes, but with caution. Those are simulated ticks and that is why it can never be exactly the same as runtime, but at least now they are much closer to reality than it used to be before