The Great HGB Foward Test

Post by **drunka** » Tue Jan 20, 2015 9:45 pm

DoctorGremlin » Today, 19:03 wrote:
Maximilian » 5 minutes ago wrote:Common Doctorgremlin,
you´re cheating. You gave your HGB a nice sunbath, a shower afterwards and you fed it after midnight. 40 open orders, 172 pending orders and 101 closed trades. What are you up to?
Sorry, no cheating here, i haven't touched both instances since the Setup on sunday.
but i gave each HGB on each chart a separate magic number, i think that might be the difference.

I thought the point was that we all have the exact ootb settings so we can compare. If we change variables the test round results become false.

Post by **trendinator** » Tue Jan 20, 2015 10:21 pm

triangla » Wed Jan 21, 2015 7:29 am wrote:Hi Guys,

A question for the more knowledgeable "Coders", Is it possible that the problem with signals is related to not having "stdlib.ex4" in the library folder?

I am running a different instance on a Mac (I don't want to screw our test) and I noticed that when I attached the expert "2o-b" to the chart it shows "stdlib.ex4" as a dependency.
In our test, we didn't install the "stdlib.ex4" in the library folder, or at least I didn't. My library folder is empty.

I am not a coder, if this question doesn't makes sense, please disregard and fill free to delete my post.

on a Mac, ha! good luck with that! are you using Wine or a vmware or what? don't expect it to work properly on a Mac, Empty4 is already glitchy enough as it is.

Post by **trendinator** » Tue Jan 20, 2015 10:23 pm

WhoKnows » Tue Jan 20, 2015 1:53 pm wrote:Haven't the time to participate now, but following with interest.

Regarding differences:
I remember few years ago when we were comparing NinjaTrader strategy performance between a group of traders, one issue turned out to be that the computer times of the participants were not the same. Some were out several minutes. At the time we used a thingy to force regular synchronisation.

It would be interesting to see if all the Amazon VPS's are more similar (as these would run at the same time stamp).

you could be onto something here. ive posted in the HGB basics thread about this.

Post by **trendinator** » Tue Jan 20, 2015 10:24 pm

DoctorGremlin » Wed Jan 21, 2015 4:03 am wrote:
Maximilian » 5 minutes ago wrote:Common Doctorgremlin,
you´re cheating. You gave your HGB a nice sunbath, a shower afterwards and you fed it after midnight. 40 open orders, 172 pending orders and 101 closed trades. What are you up to?
Sorry, no cheating here, i haven't touched both instances since the Setup on sunday.
but i gave each HGB on each chart a separate magic number, i think that might be the difference.

CHEATER! haha. wtf is this magic number sorcery?

Post by **DoctorGremlin** » Tue Jan 20, 2015 10:29 pm

The magic number is an identifier for the EA so that it knows which trades it should handle. For example, my HGB on EURUSD has magic number 2, so this HGB should only handle this trades. When you run 28 instances, and each one has magic number 0, then it might be possible that for example the HGB on EURUSD Closes trades on USDCHF.

Therefore, Magic number has nothing to do with the OOTB Settings, it simply is a (safety?) feature that each EA only trades its own pair.

At least, this is my understanding of the magic number.

Post by **trendinator** » Tue Jan 20, 2015 10:35 pm

DoctorGremlin » Wed Jan 21, 2015 8:29 am wrote:The magic number is an identifier for the EA so that it knows which trades it should handle. For example, my HGB on EURUSD has magic number 2, so this HGB should only handle this trades. When you run 28 instances, and each one has magic number 0, then it might be possible that for example the HGB on EURUSD Closes trades on USDCHF.

Therefore, Magic number has nothing to do with the OOTB Settings, it simply is a (safety?) feature that each EA only trades its own pair.

At least, this is my understanding of the magic number.

thanks for the explanation, i do actually know what the magic number is. i should rephrase my question. wtf are you you doing changing the settings!? it was supposed to be OOTB.

well actually its not that bad, you may have given us a good clue. your accounts look good at the moment on closed traded, but check out those open trades, -1500 pips or so.

triangla · Post by **triangla** » Tue Jan 20, 2015 10:37 pm

trendinator » Tue Jan 20, 2015 2:21 pm wrote:
triangla » Wed Jan 21, 2015 7:29 am wrote:Hi Guys,

A question for the more knowledgeable "Coders", Is it possible that the problem with signals is related to not having "stdlib.ex4" in the library folder?

I am running a different instance on a Mac (I don't want to screw our test) and I noticed that when I attached the expert "2o-b" to the chart it shows "stdlib.ex4" as a dependency.
In our test, we didn't install the "stdlib.ex4" in the library folder, or at least I didn't. My library folder is empty.

I am not a coder, if this question doesn't makes sense, please disregard and fill free to delete my post.
on a Mac, ha! good luck with that! are you using Wine or a vmware or what? don't expect it to work properly on a Mac, Empty4 is already glitchy enough as it is.

================

For the test, I am using my home Win 7 PC. I am not touching the Home PC until we are done with the test !
I also have a MBP for work and I am using an app called "PlayOnMac" (wine) to run Empty4.

Just to be clear, the empty library folder is on my Home PC running the test.

Post by **WhoKnows** » Tue Jan 20, 2015 10:42 pm

WhoKnows » Tue Jan 20, 2015 1:53 pm wrote:
Regarding differences:
I remember few years ago when we were comparing NinjaTrader strategy performance between a group of traders, one issue turned out to be that the computer times of the participants were not the same. Some were out several minutes. At the time we used a thingy to force regular synchronisation.

It would be interesting to see if all the Amazon VPS's are more similar (as these would run at the same time stamp).

Following up on this point.

If you have screenshots of the trades you can see how much the computer that is running is out of time sync with the broker time.
The first action would be to ask everybody to sync their computer time before starting the test.

The second point would be the HgiReadDelaySeconds setting. The larger this setting, the more chance there is that a signal that comes and disappears within the ReadDelay period is missed. And with different computers looking at different timestamps within the ReadDelay period... different computers may miss different temporary signals.
An option would be to decrease the HgiReadDelaySeconds. I did some experimenting on my computer and took it from 60 to 10 second without seeing any difference in resource use. It would take some experimenting to see how low you can go with e.g. the Amazon VPS.
But theoretically... the lower you go, the less chance that you miss temporary signals and the more chance that different computers will see the temporary signal. And the closer the result from different computers would be.

For strategy performance however there might be a conflicting reason to increase the HgiReadDelay as this decrease the chance that HGB registers a temporary signal that does not stick. We might not want temporary signals that may turn out to be incorrect tradewise.
It would be interesting to see what effect increasing the HgbReadDelay has on the performance of HGB. Maybe take it from 60 too 300 seconds to weed out bad signals .... On a 4H trading TF 5minutes is not much. Need to think this through. Just an hypothesis, no proof.

As always .... many unknowns...

Post by **trendinator** » Tue Jan 20, 2015 10:46 pm

triangla » Wed Jan 21, 2015 8:37 am wrote:
================

For the test, I am using my home Win 7 PC. I am not touching the Home PC until we are done with the test !
I also have a MBP for work and I am using an app called "PlayOnMac" (wine) to run Empty4.

Just to be clear, the empty library folder is on my Home PC running the test.

that's strange. something must've gone wrong with your installation, because this should be installed by default, no need to add it. or you're looking at the wrong folder - open Empty4 and go to File | open data folder.

triangla · Post by **triangla** » Tue Jan 20, 2015 11:15 pm

trendinator » Tue Jan 20, 2015 2:46 pm wrote:
that's strange. something must've gone wrong with your installation, because this should be installed by default, no need to add it. or you're looking at the wrong folder - open Empty4 and go to File | open data folder.

Yes, that's what I did.
Open Empty4, File=>Open Data folder=> MQL4=> Libraries=> "This folder is empty."