| Pages:
1
..
31
32
33
34
35
..
47 |
bartek
Junior Member
Posts: 40
Registered: 26-3-2020
Member Is Offline
|
|
Quote: Originally posted by admin  | | Quote: Originally posted by bartek | From your NQ/ES/YM guide:
| Quote: |
5) Walk forward all members in this group. Choose systems that have got good walk forward results for live trading.
|
Maybe it is a dummy question but I am curious.
What is the reason to run WF for all family members as they differentiate only by parameters and we optimize just parameters with WF? Isn't sufficient
to run WF only for one member? |
Good question.
You should expect somewhat different results in each wf.
The range used is relative to the initial parameters, and the WF is Genetic, so a random seed is also used in each WF.
Its also very easy to do, and not cpu or time intensive to run a number of WF. |
And what if we get failed OOS WF for one member? Then whole family is failed and better to find another or check next members result?
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
@bartek. Im not decided on that. I will be looking at such issues over time. Bottom line the methodology as described overall gave superb OOS results
on ES NQ YM, and with tiny tweak I think the Dax too.
I need to confirm the Dax as its going on what one user has sent me
|
|
|
bartek
Junior Member
Posts: 40
Registered: 26-3-2020
Member Is Offline
|
|
Peter, could you please reveal subtle differences for YM and ES as I can not achieve such good results as for NQ?
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by bartek | | Peter, could you please reveal subtle differences for YM and ES as I can not achieve such good results as for NQ? |
I cant remember whats better, es or nq.
Only change I make is start date on es make 1997
you could also try 2000 vs 2007, but I think 1997 was best.
Thats going from memory, not sold facts in front of me
|
|
|
bartek
Junior Member
Posts: 40
Registered: 26-3-2020
Member Is Offline
|
|
| Quote: Originally posted by admin | | Quote: Originally posted by bartek | | Peter, could you please reveal subtle differences for YM and ES as I can not achieve such good results as for NQ? |
I cant remember whats better, es or nq.
Only change I make is start date on es make 1997
you could also try 2000 vs 2007, but I think 1997 was best.
Thats going from memory, not sold facts in front of me |
Thanks. Have you tried tertiary filter with SF CloseLessPrevCloseDBpv?
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by bartek | | Quote: Originally posted by admin | | Quote: Originally posted by bartek | | Peter, could you please reveal subtle differences for YM and ES as I can not achieve such good results as for NQ? |
I cant remember whats better, es or nq.
Only change I make is start date on es make 1997
you could also try 2000 vs 2007, but I think 1997 was best.
Thats going from memory, not sold facts in front of me |
Thanks. Have you tried tertiary filter with SF CloseLessPrevCloseDBpv? |
No, I've done nothing with TF, but would be very interested in the results.
There seems to be infinite things to try in gsb, but the direction we are going in is great 
|
|
|
bartek
Junior Member
Posts: 40
Registered: 26-3-2020
Member Is Offline
|
|
| Quote: Originally posted by admin | | Quote: Originally posted by bartek | | Quote: Originally posted by admin | | Quote: Originally posted by bartek | | Peter, could you please reveal subtle differences for YM and ES as I can not achieve such good results as for NQ? |
I cant remember whats better, es or nq.
Only change I make is start date on es make 1997
you could also try 2000 vs 2007, but I think 1997 was best.
Thats going from memory, not sold facts in front of me |
Thanks. Have you tried tertiary filter with SF CloseLessPrevCloseDBpv? |
No, I've done nothing with TF, but would be very interested in the results.
There seems to be infinite things to try in gsb, but the direction we are going in is great |
True. And I found I have much fun improving results
|
|
|
REMO755
Member
Posts: 181
Registered: 11-4-2021
Member Is Offline
|
|
Hello,
I get lost with secondary and tertiary filters.
To understand ..... What does the tertiary filter do in RSI?
Where is the code for the indicators?
|
|
|
portfolioquanttrader2020
Member
Posts: 269
Registered: 30-4-2020
Member Is Offline
|
|
I have doubts about the data configuration:
- What nomenclature should I use for the east coast of the United States, with data of 1 minute and that include the schedule of two days, from 18
hours today, which is the entrance, until 3 or 4 in the morning of the day following? How to name the data?
- In the previous case, how should I configure contracts list, because the one I have from the ES has a session time from 830 to 1500?
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by REMO755 | Hello,
I get lost with secondary and tertiary filters.
To understand ..... What does the tertiary filter do in RSI?
Where is the code for the indicators? |
I have not yet used TF at all, so best wait till I know how its best used.
The code will be the same indicators, used in a different way.
(slightly different architecture. )
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by portfolioquanttrader2020 | I have doubts about the data configuration:
- What nomenclature should I use for the east coast of the United States, with data of 1 minute and that include the schedule of two days, from 18
hours today, which is the entrance, until 3 or 4 in the morning of the day following? How to name the data?
- In the previous case, how should I configure contracts list, because the one I have from the ES has a session time from 830 to 1500?
|
I think in the past, I showed similar setup for HG. Adjust that.
you can put any session time against any symbol. ie 24x7
It would be better that you use the nq/es/ym settings defaults as this works well,
and what your are trying to do is much harder, and less likely to work
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
here is an update on fitness for NQ
Still doing more tests, and there are 4 new fitness types in 62.38 released today
|
|
|
NickW
Junior Member
Posts: 29
Registered: 12-2-2020
Member Is Offline
|
|
Hi Peter,
The number for BC Avg of 4.05, what is that the average of? I know it's the avg of B and C stats, but what is in B and C?
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by NickW | Hi Peter,
The number for BC Avg of 4.05, what is that the average of? I know it's the avg of B and C stats, but what is in B and C?
|
If we change fitness in all our tests, gsb is using np/dd for all metrics here as a second fitness
bc ave + ave b + ave c just as bce is ave b ave c + ave c
spreadsheet sorts on highest bce ave being first
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
I hope to have the results of what fitness works best on NQ. < 1 day of testing to go. If Anyone has more workers to contribute, please send share
keys
thanks for those that have so far
You can see our traditional fitness np/dd and np*at are we'll down the list
Thanks received (4):
+1 OUrocketman at 2021-05-12 15:31:45 +1 Siem at 2021-05-12 10:27:00 +1 LucaRicatti at 2021-05-12 07:22:58 +1 Carl at 2021-05-12 06:57:52
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Here is the final result
Whats good is mfe, NPlateentry, NP and AveTrade are at the top consistently.
Im doing similar testing on ES long only swing systems now.
Thanks for those who contributed workers for this project.
If you contribute workers, you will also get my setup for ES long only systems on your workers, which you can copy
Thanks received (1):
+1 Piet at 2021-05-13 23:35:44
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Next build of GSB will have have nq data file, nq macros for building nq/es/ym systems
The only change from previous macro is fitness is no longer np/dd but np.at.mea.nlateEntry
whats VIP is the data file has data exported as ticks (on your ts/mc chart = up/down volume)
if volume was exported its up volume only. This is critical to get right as many nq systems use indicators that use volume
Thanks received (3):
+1 sfuser108 at 2021-05-14 09:10:01 +1 Siem at 2021-05-14 05:58:51 +1 Bruce at 2021-05-13 22:54:02
|
|
|
rws
Member
Posts: 114
Registered: 12-6-2017
Member Is Offline
Mood: No Mood
|
|
@peter
Very interesting issue.
If you optimize on Netprofit x MAE x MFE, the result with higher MAE has a better fitness. That should not be. It should be Netprofit/MAE x MFE
You want to have a small MAE so it should be under the /. Or rather have a exponent
-1 for dividing. A^-1 = 1/A.
I would love to see what is happening if the number of bars is possible.
For example (NP x AT)/(MAE x numberoffbars)
That would rate trades with less bars higher with less maximum adverse excursion or rather average adverse excursion.
Or rather have any other combination by a custom formula /numberofbars user configurable.
I have had good results with /numberofbars in a fitness function in other optimizers especially if trades go overnigt where this is very important. I
really miss that in GSB. You typically see optimizers have very long trades
with big MAE if PF x win% should be high because then these numbers are better if there is no /numberofbars. But that is fake performance to get a
straight line. The profit divided by the time or bars in the market also tends to find better systems OOS is my experience.
| Quote: Originally posted by NickW | Hi Peter,
The number for BC Avg of 4.05, what is that the average of? I know it's the avg of B and C stats, but what is in B and C?
|
|
|
|
rws
Member
Posts: 114
Registered: 12-6-2017
Member Is Offline
Mood: No Mood
|
|
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
@rws. Im very open to your ideas of number of bars. Im going to think seriously on that.
all fitness's are adjusted so whats best is positive. ie mae will be 1/mae
|
|
|
portfolioquanttrader2020
Member
Posts: 269
Registered: 30-4-2020
Member Is Offline
|
|
Hi peter
I need to know how to interpret these graphs.
I see that you do less trades in WFO AND WF CURRENT than in the in sample period. What is the reason?
|
|
|
portfolioquanttrader2020
Member
Posts: 269
Registered: 30-4-2020
Member Is Offline
|
|
Hi peter
I need to know how to interpret these graphs.
I see that you do less trades in WFO AND WF CURRENT than in the in sample period. What is the reason?
 
|
|
|
portfolioquanttrader2020
Member
Posts: 269
Registered: 30-4-2020
Member Is Offline
|
|
Hi peter
These curves look like they look good.
What do you think?
|
|
|
portfolioquanttrader2020
Member
Posts: 269
Registered: 30-4-2020
Member Is Offline
|
|
Hi peter
These curves look like they look good.
What do you think?
|
|
|
bartek
Junior Member
Posts: 40
Registered: 26-3-2020
Member Is Offline
|
|
How did you achieve the same trades count for all curves? For all my results with NP*AT*MFE*NPLateEntry I get significantly less trades for CurrentWF.
I tried to set commision to 10 or 20 but without improvement for trades count differences.
|
|
|
| Pages:
1
..
31
32
33
34
35
..
47 |
![[*]](images/davis/default_icon.gif){kind=icon}
