Optimisation fail

Viewing 13 posts - 1 through 13 (of 13 total)

Author

Posts
04/02/2018 at 10:17 PM #66960 quote

irioton
Participant

Average

Hi all ! I’ve got some difficulties with the plateforme, I want to optimise 2 variables (“m” and “d” in the picture), from 1 to 100, the others are fixed.

“d” is tested, but “m” stays to 1. Did someone have that before ?

Sans-titre.png

04/02/2018 at 10:55 PM #66964 quote

Vonasi
Moderator

Master

Normally it is because you have the value still fixed somewhere in your code so that it has no effect on the results. Either that or it is not used in the code at all.

04/02/2018 at 11:16 PM #66965 quote
irioton
Participant

Average
here is the code, both variables are used :
```
//trailing stop function
trailingstart = d //trailing will start @trailinstart points profit

clow = (lowest[m](low))-tradeprice
chigh= tradeprice-(highest[m](high))

//reset the stoploss value
IF NOT ONMARKET THEN
newSL=0
ENDIF
//manage long positions
IF LONGONMARKET THEN
//first move (breakeven)
IF newSL=0 AND close-tradeprice(1)>=trailingstart*pipsize THEN
newSL = tradeprice(1)+1
else
SET STOP ptrailing 83
ENDIF
//next moves
IF newSL>0 AND clow>trailingstart then
newSL = newSL+clow*pipsize
ENDIF
ENDIF
 
//manage short positions
IF SHORTONMARKET THEN
//first move (breakeven)
IF newSL=0 AND tradeprice(1)-close>=trailingstart*pipsize THEN
newSL = tradeprice(1)-1
else
SET STOP ptrailing 83
ENDIF
//next moves
IF newSL>0 AND chigh>trailingstart then
newSL = newSL-chigh*pipsize
ENDIF
ENDIF
 
//stop order to exit the positions
IF newSL>0 THEN
SELL AT newSL STOP
EXITSHORT AT newSL STOP
ENDIF
```
04/03/2018 at 9:29 AM #67002 quote

GraHal
Participant

Master

Might sound daft but it can mean that ‘m’ at value 1 gives the highest gain? Make m value fixed at 5 and see if gain reduces in the optimiser results?

What is your range of values for m?

PS / Edit~
You may need a range with a much higher max value for d and m to get a optimised value other than 1 for m ??

04/03/2018 at 9:49 AM #67003 quote

Nicolas
Keymaster

Master

You should not multiply clow and chigh by pipsize at lines 21 and 35. Please try without.

04/03/2018 at 11:27 AM #67017 quote

irioton
Participant

Average

@GraHal : “d” and “m” are both tested from 1 to 100 with a step of 1

@Nicolas : I did what you suggested, it doesn’t change the result, I still got an error message, but now it comes before the end of optimisation…

04/03/2018 at 11:30 AM #67018 quote

irioton
Participant

Average

I sent a message to prt, they said they’re going to look into it…

04/03/2018 at 12:12 PM #67023 quote

Nicolas
Keymaster

Master

What ‘error message’ !?

Are you sure your ‘m’ variable is not set to a fixed value into the variables optimizations window?

04/03/2018 at 1:00 PM #67025 quote

GraHal
Participant

Master

If you put your full code on here then I will try it the optimiser for you?

It is probably finger trouble / something daft that we have all done before also so we may spot it in your set up where we couldn’t see it when we made the same error!!! 🙂 🙂

04/03/2018 at 1:43 PM #67028 quote

irioton
Participant

Average

@Nicolas : yes i’m sure the values are not set to fixed. The error message just says there is an error, but not which one, that’s why I wrote to PRT.

@GraHal : attached is the code untouched, yes could you optimise it for me please ?

I would be glad if the error was mine, maybe you can find what’s wrong by testing it yourself.

This is Dax 1 min.

dax-moyenne-mobille-1mtrail.itf

04/03/2018 at 4:40 PM #67041 quote

GraHal
Participant

Master

GraHal wrote:

Might sound daft but it can mean that ‘m’ at value 1 gives the highest gain?

Apologies for delay, had to take my car for annual MOT! 🙂

It’s what I said above.

See attached for 10k bars and m is at different values
See attached for 100k bars and m stays at vaue = 1
See attached for 100k bars for m range 10 to 250 in step of 10 and m is at different values.

Cheers
GraHal

irioton thanked this post

10k-bars.jpg 100k-bars.jpg 100k-bars-10-to-250.jpg

04/03/2018 at 11:52 PM #67081 quote
irioton
Participant

Average
Thank you so much ! So I was worrying for nothing…This made me think about several things :
- I can’t see the other possibilities in the report windows when I do the tests, is that normal ?
- DO you have an error message too ?
- my trailing highs/lows snippet doesn’t work, I have to work on it :s
04/04/2018 at 10:07 AM #67111 quote

GraHal
Participant

Master

irioton wrote:

I can’t see the other possibilities in the report windows when I do the tests, is that normal ?
Yes I think we only see 100 results.

DO you have an error message too ?
No I got no error message but I didnt leave it running until it finished.

My trailing highs/lows snippet doesn’t work, I have to work on it :s
I agree.

Rem out Nics stop and substitute a normal pstop and ptrailing to get a quick idea what values would work … I did and it was like 240 and 130 … and even then overall gain was very low, but maybe I didn’t leave it running until optimiser completed?

Just an idea I use sometimes
GraHal
Author

Posts