Download MetaTrader 5

ATC Rules not so specific.

To add comments, please log in or register
Your conversation partner speaks foreign language? Automatic translation will help you!
Joaquin Poudereux
489
Joaquin Poudereux 2010.09.26 15:27 

ATC Rules didn't say in anyway that the EA should finish the backtest in less than 15 minutes. The rules only said "Be efficient in CPU resources and computer memory".

For next contests, I suggest to explain better the rules.

You can imagine my big deception when I received the notification that my ea didn't finish the tests, after all the effort of porting it from mql4, and learning mql5 in 15 days.


I also would thank a lot one article in your site about optimizing memory/cpu in our EAs.

Anyway, it's been a rough path to write my ea, but i'm proud of it..


Rgards

rforex
15
rforex 2010.09.26 18:15  
yokinfx:

ATC Rules didn't say in anyway that the EA should finish the backtest in less than 15 minutes. The rules only said "Be efficient in CPU resources and computer memory".

For next contests, I suggest to explain better the rules.

You can imagine my big deception when I received the notification that my ea didn't finish the tests, after all the effort of porting it from mql4, and learning mql5 in 15 days.


I also would thank a lot one article in your site about optimizing memory/cpu in our EAs.

Anyway, it's been a rough path to write my ea, but i'm proud of it..


Rgards

In this case, the earlier tester should have rejected the EA and not after the registration is closed and certainly not when the first tester passed the SAME EA with flying colors! Mine passed their test initially and failed their repeat test after the registration close, in which I could not do anything. I agree it was a big deception
Dumbledore
114
Dumbledore 2010.09.26 20:24  
They may have tested a wrong EA before the registration close. I actually suspected this when I see roughly a same result with the one I uploaded before (with a lower leverage) and sent a private message to them. No reply and no action. I didn't chase because I thought everyone must be busy at that time. Also my EA was actually finished way earlier and passed numerous test and I believed there should indeed be no problem this time. In my case I had a typo on the final parameter set, making the EA sending a wrong volume in only one trade attempt during the back test.
MQL5.community - User Memo
  • 2010.02.25
  • MetaQuotes Software Corp.
  • www.mql5.com
You have just registered and most likely you have questions such as, "How do I insert a picture to my a message?" "How do I format my MQL5 source code?" "Where are my personal messages kept?" You may have many other questions. In this article, we have prepared some hands-on tips that will help you get accustomed in MQL5.community and take full advantage of its available features.
ranel
5
ranel 2010.09.26 23:15  

I'm also a victim of the 15 min hidden policy. I can see why MQ had to put some time out limit for their automatic testing, especially when expecting close to 1800 EA tests.

However, this 15 minutes limit had to be specified in rules!

When MQ opened up the door to more extensive EAs with the new MQL5 and when encouraging to develop multi currency EAs, they should expect more complex EAs that compute more and the time limit should be adjust accordingly or else they will get more of the same kinds of EAs.

My EA finished the qualification test in 19 minutes on my machine and I could easily adjust it to profitably terminate in less than 15 minutes had I known this restriction.

MetaQuotes organizers,  a reasonable and fair solution in this case is to either retest those slower EAs with looser timeout restriction (say, 30 minutes) or give those EAs creators a short time (say a day) to fit the 15 minutes restriction. Idealy the time out restriction should be a proportinal to the number of pairs used by the EA, not to mention that in many cases the loading and synchronization time required for multiple symbols in the tester is outrageous. 

 

 

 

 

 

Atsushi Yamanaka
5418
Atsushi Yamanaka 2010.09.27 02:10  
I got same one too..
My latest EA have more than 50000 trades in 7 month, got about 2.5 million on my test.
No way to finish the test in 15 min.
If I knew the hidden rule, I could make it back to the 11 symbols version like I got OK on your test on the day ” Phantom dead line".

The one before took 7 min 50 sec to get 1.25 mil. with 29594 trades.

So the last one would take 15 min and 40 sec on your machine if twice money takes twice time.

This 40 sec is the reason I got be OUT?

 

2010.09.22 15:21

1. Copy
   experts\contest2010.ex5 ok

2. Compile

3. Configuration
   contest2010 on EURUSD:H1 every tick 2010.01.01-2010.08.01


4. Start
   finished in 7 min 50 sec


5. Statistics
   1392 kb of log files
   29594 trades, 59210 deals, profit 1250429.03 USD

No errors!

If you didn't have the trouble with your testing machine and didn't extend the dead line, I didn't add the 12th symbol....orz

Just I wish is to get back to the FIRST DEAD LINE!

2010.09.24 04:42

1. Copy
   experts\contest2010-ver2.ex5 ok

2. Compile

3. Configuration
   contest2010-ver2 on EURGBP:H1 every tick 2010.01.01-2010.08.01


4. Start
   expert takes too long time (more than 15 minutes)

1 errors
Johnny
140
Johnny 2010.09.27 02:54  

Hi guys, be patient, i believe MetaQuotes staff will look at this issue and approve manually .

my opinion , on alohafx's case, it  will not be fair for him if he can't participate atc2010, cuz he spent time prepared his ea and got accepted on the first deadline but on yr extend the dead line ( for unprepared ppl to join)  , he uploaded new version, unfortuately get into below issue which he didn't aware of.

expert takes too long time (more than 15 minutes)


rforex
15
rforex 2010.09.27 03:26  
With 6 more days to go, I really hope ATC team will look into this problem. It is really unfair because they provided false assurance during their initial testing. I agree that with MT5 and the ability to utilize complex algorithms, more time is needed for computation, especially backtesting multi currencies, and ATC should not penalize us for utilizing that.

rforex
15
rforex 2010.09.27 09:05  

Again the ATC team rejected my EA with the same reason:

expert takes too long time (more than 15 minutes)

Because of this problem, I disabled backtesting of other currencies other than the chart it is tested on but it will run on multicurrency on forward test.

How can ATC team repetitively reject the same ex5 file that was passed before the registration closing? 

Atsushi Yamanaka
5418
Atsushi Yamanaka 2010.09.27 22:09  
THANK GOD! THANK stringo! THANK MetaQuotes, all sponsors and all jury.
THANK yster, rforex and everyone here.

Now I got resurrection!

I love all you guys and I wish all other people here will be IN.
GOOD LUCK for everybody!


traderobo
4
traderobo 2010.09.27 23:46  

Metaquotes should also look at this:


profit is too small

I submitted my EA  where i put all my effort in to implement a strategy that is based on statistical info that only would work at the last 3 months of the year, during the championship. It is normal that my EA is not profitable in other time periods. The test is complete error free, the only automated reason for disqualification Metaquotes give is: "Profit is too small".

I think this is a discutable reason in general, because backtest are not same as live trading, and profitable EA's can have a period of loss in for instance a decade time, but on top of that: For my EA, Metaquotes is testing in a wrong time period, the time period of their test report is 2010.01.01-2010.08.01

For the research i did, i know my EA works perfect in the period where it is made for, all my effort for nothing just because of reason "Profit is too small"?

I think it is unfair if the ATC team threats the participants like this, we have spent sweat and tears to get a working and error free EA especially to compete in this competition. 

I would kindly ask Metaquotes to overthink this.

Anyone have a suggestion who i can contact about my issue, because i already tried though the message option in my account, but did not get a personal answer.

ranel
5
ranel 2010.09.27 23:55  
traderobo:

Metaquotes should also look at this:

I submitted my EA  where i put all my effort in to implement a strategy that is based on statistical info that only would work at the last 3 months of the year, during the championship. It is normal that my EA is not profitable in other time periods. The test is complete error free, the only automated reason for disqualification Metaquotes give is: "Profit is too small".

I think this is a discutable reason in general, because backtest are not same as live trading, and profitable EA's can have a period of loss in for instance a decade time, but on top of that: For my EA, Metaquotes is testing in a wrong time period, the time period of their test report is 2010.01.01-2010.08.01

For the research i did, i know my EA works perfect in the period where it is made for, all my effort for nothing just because of reason "Profit is too small"?

I think it is unfair if the ATC team threats the participants like this, we have spent sweat and tears to get a working and error free EA especially to compete in this competition. 

I would kindly ask Metaquotes to overthink this.

 Anyone have a suggestion who i can contact about my issue, because i already tried though the message option in my account, but did not get a personal answer.

 

Your statistical approach to this sounds really nice and valid.  I really don't understand the MQ attitude too. The rules of any competition, including the qualifying, must be transparent.  I tried to contact one of the organizers but I got a one liner stating that I was too late nad the Mantra of "your EA must be efficient".

I don't believe any objective mind will agree that pulling hidden qualification rules is fair. Have you tried contacting the Jury? 

12
To add comments, please log in or register