Self learning expert

New comment
 

Found this on Poul Trade Forum at www.kbpauk.ru the first backtest profit factor .99 2nd,3rd, and fourth backtest better this one attached is the 5th backtest with a profit factor 4.00+ and less than 3%drawdown for this year.Tried fwd. testing this week with no trades. Seems it doesn't work on fwd. demo. Could somebody please look at this and figure out what is wrong. From what I understand this ea. looks at its own history files and make decisions on this.

Thanks in advance,

mrtools

 

Maybe the fact that it relies on history to generate trades is the reason it doesn't work in real time. I'm unable to get it to backtest for me for some reason, but try changing the setting "Read History" to 0 and then backtest it, and post your results.

 

Self learning expert

Just remembered when I first tested this expert got a error message " invalid file handle", I turned metatrader off and then back on and then it backtested. Got the same error message when putting it on the charts, did the same, no more error message but no trades.For some reason it doesn't seem to read the files when live.

regards,

mrtools

 

History to zero

Hi Aegis

This expert will not work with history on zero due to the fact it relies on history to make future decisions as far as I understand the code.

regards,

mrtools

 
mrtools:
Found this on Poul Trade Forum at www.kbpauk.ru the first backtest profit factor .99 2nd,3rd, and fourth backtest better this one attached is the 5th backtest with a profit factor 4.00+ and less than 3%drawdown for this year.Tried fwd. testing this week with no trades. Seems it doesn't work on fwd. demo. Could somebody please look at this and figure out what is wrong. From what I understand this ea. looks at its own history files and make decisions on this.

Thanks in advance,

mrtools

mrtools,

Please give the full link to the page of that forum where you found this EA (i may read something and may be will find some important information to post here).

 

I think it has to build a history first. When I first ran it it ran at a normal back test speed. Untill I got the error message. Then I turned meta off and back on. Now it runs in slow motion, even though I have it at full speed. It took 7 days befor it took a trade in the back test. But since it took the first trade it is taking a trade almost every day. So running it live might take some days or week or two befor it makes a trade.

I expect as the history grows it will possibly trade more often and more accurate.

Doesn't look to just read the history in meta didn't it have to creat the file to read? Since we got that error befor restarting it? It only losses 25 pips at a time. If I can get good back test I might try to run it on real account with it set to micro lots.

 
xxDavidxSxx:
I think it has to build a history first. When I first ran it it ran at a normal back test speed. Untill I got the error message. Then I turned meta off and back on. Now it runs in slow motion, even though I have it at full speed. It took 7 days befor it took a trade in the back test. But since it took the first trade it is taking a trade almost every day. So running it live might take some days or week or two befor it makes a trade.

I expect as the history grows it will possibly trade more often and more accurate.

Doesn't look to just read the history in meta didn't it have to creat the file to read? Since we got that error befor restarting it? It only losses 25 pips at a time. If I can get good back test I might try to run it on real account with it set to micro lots.

hi dave

it is the first time that i see an EA correct itself and the BT result each time i run it.i'm backtesting it from 2006.01.01 at the first time the result is awefull but at the moment at the 2nd time it does much better.i am doing it with real data.i think we can back test it more and more and find a good result then we can put the files that are(4 eurusd):

GS_GS_EURUSD

last_buyEURUSD60

last_sellEURUSD60

in the metattraer\files\ folder.i think by this way we dont need to wait long time to EA make the history files.

i'll post the BT results and made files soon.

thx

 
mrtools:
Found this on Poul Trade Forum at www.kbpauk.ru the first backtest profit factor .99 2nd,3rd, and fourth backtest better this one attached is the 5th backtest with a profit factor 4.00+ and less than 3%drawdown for this year.Tried fwd. testing this week with no trades. Seems it doesn't work on fwd. demo. Could somebody please look at this and figure out what is wrong. From what I understand this ea. looks at its own history files and make decisions on this.

Thanks in advance,

mrtools

I found good evaluation of this EA on alpari forum. Very good and veru sceptical evaluation here http://forum.viac.ru/viewtopic.php?p=74688&highlight=gselector#74688

besides there is some information on Pouls forum as well http://forex.kbpauk.ru/showflat.php/Cat/0/Number/117828/an/0/page/1#Post117828

Members may translate from Russian using online translator as those discussion on aplari forum is very good.

 

hi

ok here it is.this is the result of backtest(1st time)all made files and result and analysisi are attached.

Files:
first_time.rar  180 kb
 

and here the is the second time result?!?!?!?!?

Files:
sec_time.rar  190 kb
 

I am wondering...

The first test stores the info. Then you run the same test and it "knows" what the price is going to do because it has seen it do it befor.

Mabe like us taking the history and going back in time and trading the last year off of what we have seen it do.

Is it learning or mearly tricking us?

Mabe we need to run tests on just 2004-2005 and then once it does good, run it on 2005-2006 and see if it really learned anything from 04-05.

we cannt let it lose 07-08 then go back and re live the past.....lol

12345
New comment
Reason: