| Pages:
1
..
36
37
38
39
40
..
98 |
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Moresi552, im not clear. Do you want the settings from a worker to be applied back to a manager?
This can be done via settingsfullmacro. There is an option to pull settings. It wont pull in the contracts used. This has do be done manually
Thanks received (1):
+1 moresi522 at 2020-05-18 07:33:05
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
Why is even your settings showing on public cloud workers? is there a reason for this?
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Its the simplest way for GSB to function, and if there is something to be diagnosed on a worker, then we need to see the settings
|
|
|
moresi522
Junior Member
Posts: 50
Registered: 31-1-2020
Member Is Offline
|
|
Why use UseWfParams in Macro M3?
Hi,
I read macro m3 and if I'm not wrong for stats B it compares data generated from A (
Global dates 2015,useWFcurrparms false) using the same date with UseWFCurrParams = True
I don't understand the meaning of this task. Why UseWFCurrParms = true? which informations give us comparing these two data?
I would understand better if I used, for example, the same data for A and setting nth from no trade to trade to compare the IS and OOS data.
Thanks
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
Quote: Originally posted by moresi522  | Why use UseWfParams in Macro M3?
Hi,
I read macro m3 and if I'm not wrong for stats B it compares data generated from A (
Global dates 2015,useWFcurrparms false) using the same date with UseWFCurrParams = True
I don't understand the meaning of this task. Why UseWFCurrParms = true? which informations give us comparing these two data?
I would understand better if I used, for example, the same data for A and setting nth from no trade to trade to compare the IS and OOS data.
Thanks |
Hi Moresi, not sure i totally understand your question, but if i understand your correctly A and B for example gives you stats for A without WF and B
with... if your process dont value with and without WF stats.. you can use Peters market validation macro, that gives you IS relative to OOS
degradation.. if you follow peters methodology for cl then you dont use nth, just all data for build.
Thanks received (1):
+1 moresi522 at 2020-06-02 02:50:19
|
|
|
moresi522
Junior Member
Posts: 50
Registered: 31-1-2020
Member Is Offline
|
|
| Quote: Originally posted by Daniel UK1 |
Hi Moresi, not sure i totally understand your question, but if i understand your correctly A and B for example gives you stats for A without WF and B
with... if your process dont value with and without WF stats.. you can use Peters market validation macro, that gives you IS relative to OOS
degradation.. if you follow peters methodology for cl then you dont use nth, just all data for build. |
Hi Daniel,
I would like to understand the meaning of this choice. Don't just do it because you do it. What information does IS results compare to the same data
with UseWfCurrParms?
Thank you
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
| Quote: Originally posted by moresi522 | | Quote: Originally posted by Daniel UK1 |
Hi Moresi, not sure i totally understand your question, but if i understand your correctly A and B for example gives you stats for A without WF and B
with... if your process dont value with and without WF stats.. you can use Peters market validation macro, that gives you IS relative to OOS
degradation.. if you follow peters methodology for cl then you dont use nth, just all data for build. |
Hi Daniel,
I would like to understand the meaning of this choice. Don't just do it because you do it. What information does IS results compare to the same data
with UseWfCurrParms?
Thank you |
Hi Moresi,
I can just answer for myself, how i see it, not anything else.
The stats for with and without wf, gives me really no other view than is WF better to do, or not.. I use it to see if WF is worse, not anything else,
since WF is part of my process just to see how stable parameters are, nothing else. So as long as results is not worse, i will use WF. Macro is also
very good because these stats allowed you to chose what values you want to use for your own stats in excel, based on these. I for example use values
in B D F H for further analysis and comparisons. You might want to use A C E G for your further analysis or you perhaps you decide to look at only d f
and h if this is your oos..
For markets and builds where i use Nth and not all data to build on, i dont use this specific WFmacro until absolute final part in my process, it has
less value for me and need other stats to deciding things before this stage.
This is just my very personal way of using these tools.
Thanks received (2):
+1 LucaRicatti at 2020-06-02 05:32:46 +1 moresi522 at 2020-06-02 03:56:21
|
|
|
philhm
Newbie
Posts: 1
Registered: 26-7-2019
Member Is Offline
|
|
Machine Resource Settings
Hi All,
Does anyone have an idea of what would be the best configuration settings for Workers/Manager for these types of machines:
16x2 (32 Threads) Xeon 176GB
24x2 (48 Threads) Xeon 192GB
Just did an upgrade of both from 128 and trying to get them as optimized as possible for GSB.
Phil
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by philhm | Hi All,
Does anyone have an idea of what would be the best configuration settings for Workers/Manager for these types of machines:
16x2 (32 Threads) Xeon 176GB
24x2 (48 Threads) Xeon 192GB
Just did an upgrade of both from 128 and trying to get them as optimized as possible for GSB.
Phil |
Hi Phil,
let RM decide on the amount of workers.
set your cache settings from 600 /1800 seconds under worker machine resources
Thanks received (1):
+1 philhm at 2020-06-14 10:30:05
|
|
|
matchhanson
Junior Member
Posts: 16
Registered: 13-12-2017
Member Is Offline
Mood: No Mood
|
|
i am having trouble updating to the newest version (20200525.2) of RM. it tries to download the .zip file then the status changes to stopped then the
window closes, all in the space of a second. does anyone have any suggestions?
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by matchhanson | | i am having trouble updating to the newest version (20200525.2) of RM. it tries to download the .zip file then the status changes to stopped then the
window closes, all in the space of a second. does anyone have any suggestions? |
ive seen that before, but dont have logs to identify the problem
Im happy to fix if you email me teamviewer details.
I want to get the logs and fix problem for all users, not just you.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by philhm | Hi All,
Does anyone have an idea of what would be the best configuration settings for Workers/Manager for these types of machines:
16x2 (32 Threads) Xeon 176GB
24x2 (48 Threads) Xeon 192GB
Just did an upgrade of both from 128 and trying to get them as optimized as possible for GSB.
Phil |
see https://trademaid.info/forum/viewthread.php?tid=92#pid1347
you can also have cpu settings in RM high.
ram you can go lower, but go too low you may crash the machine.
ram high 15 / ram low 5gb is the extreme I would go.
A bit more buffer on ram better.
Thanks received (1):
+1 philhm at 2020-06-14 10:29:48
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
Wanted to start to use PA pro.. but having some issues getting my strats to PA, tried ELD, that comes with PA, but that cant compile in MC... tried
Randy version but that causes errors... and i can not manually save excel and convert to xml etc.. not a good approach... does anyone have a ELD
that Multichart can work with that will output strategy name and locate end file at a specified folder ?
Thanks
|
|
|
RandyT
Member
Posts: 123
Registered: 5-12-2019
Location: Colorado, USA
Member Is Offline
|
|
| Quote: Originally posted by Daniel UK1 | Wanted to start to use PA pro.. but having some issues getting my strats to PA, tried ELD, that comes with PA, but that cant compile in MC... tried
Randy version but that causes errors... and i can not manually save excel and convert to xml etc.. not a good approach... does anyone have a ELD
that Multichart can work with that will output strategy name and locate end file at a specified folder ?
Thanks |
Daniel, most recent version attached. Tell me if that fixes your issues.
Attachment: Login to view the details
Thanks received (1):
+1 Daniel UK1 at 2020-06-11 09:24:18
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
Many Thanks Randy, will test now.
Btw what is the reason you opted for a separate signal instead of a function used on the trading strategy that pulls name directly from strategy? just
curious.
|
|
|
bizgozcd
Junior Member
Posts: 53
Registered: 27-5-2020
Member Is Offline
|
|
Greetings all. I would like to test variables in standard indicators. Is there a way to get at the indicator settings or do I need to set up custom
indicators to do this? Thank you!
|
|
|
Carl
Member
Posts: 342
Registered: 10-5-2017
Member Is Offline
Mood: No Mood
|
|
Welcome bizgozcd,
I think this is the only way to manually control the parameter settings:
First you build some strategies in GSB.
Then right click a strategy, then choose "Run walk-forward", then "AI WF (parameter analysis)".
The results can be seen under tab "AI WF".
You can set the parameter ranges in the left window under "AI params".
Thanks received (1):
+1 bizgozcd at 2020-06-16 17:12:08
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
Hi,
I believe there is an issue when it comes to creating a macro for settings (tried both full and partial etc), this is the case when i load an older
saved opt setting from a older manager version , in a never manager.
I my attempt i loaded an old opt setting into the new manager, changed the setting in the file meny (adapt to new settings and default) to NO, i then
got my old setting without new defaults (i believe), and then tried to create a macro to share to another user, however.. the settings that the macro
delivered was not at all what was in my saved opt setting, so something fishy here.
Another issue, is that when you open a new manager, and try to load an old opt setting, set the switch to NO for (adapt to new settings and
defaults).. you get your old settings... but if i then switch back to YES (adapt to new settings and defaults) i get neither standard defaults or the
old saved settings.
Both these issues makes me worried and not comfortable that this feature really gives me for sure my old settings only in a new manager and handles
this properly.
Is there something i misunderstand?
I believe the correct way is to load the old opt setting, and then switch to NO for (adapt to new settings and defaults), and not switch to NO before
load? (however when trying this after the load of the opt setting, the swtich is back to YES..
Some confusion on this from my side
Cheers
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by Daniel UK1 | Hi,
I believe there is an issue when it comes to creating a macro for settings (tried both full and partial etc), this is the case when i load an older
saved opt setting from a older manager version , in a never manager.
I my attempt i loaded an old opt setting into the new manager, changed the setting in the file meny (adapt to new settings and default) to NO, i then
got my old setting without new defaults (i believe), and then tried to create a macro to share to another user, however.. the settings that the macro
delivered was not at all what was in my saved opt setting, so something fishy here.
Another issue, is that when you open a new manager, and try to load an old opt setting, set the switch to NO for (adapt to new settings and
defaults).. you get your old settings... but if i then switch back to YES (adapt to new settings and defaults) i get neither standard defaults or the
old saved settings.
Both these issues makes me worried and not comfortable that this feature really gives me for sure my old settings only in a new manager and handles
this properly.
Is there something i misunderstand?
I believe the correct way is to load the old opt setting, and then switch to NO for (adapt to new settings and defaults), and not switch to NO before
load? (however when trying this after the load of the opt setting, the swtich is back to YES..
Some confusion on this from my side
Cheers |
Hi Daniel
im happy to look at the macro with you over team viewer in 14 or so hours onwards, else monday
I will ask the programmer on the other issue.
|
|
|
bizgozcd
Junior Member
Posts: 53
Registered: 27-5-2020
Member Is Offline
|
|
Question. I completed a run that generated 8200 unique systems. When I ran Macro 1, it only added to favorites and did walks forward on 12 of them. I
repeated it and it did the same thing. Any idea why this may be happening? I verified in the macro that it is set to 250. Thanks again.
|
|
|
NickW
Junior Member
Posts: 29
Registered: 12-2-2020
Member Is Offline
|
|
Edit the macro and see what the criteria is.
Normally it will filter your 8200 systems by PF > 1.8 and Num Trades > 100 and Pearson > 0.95 and then from the remaining systems that passes this
test, it will select the Top250.
Sometimes you just dont have enough systems that passes the criteria to select 250.
|
|
|
bizgozcd
Junior Member
Posts: 53
Registered: 27-5-2020
Member Is Offline
|
|
| Quote: Originally posted by NickW | Edit the macro and see what the criteria is.
Normally it will filter your 8200 systems by PF > 1.8 and Num Trades > 100 and Pearson > 0.95 and then from the remaining systems that passes this
test, it will select the Top250.
Sometimes you just dont have enough systems that passes the criteria to select 250. |
Aha! Thank you. Turns out it was working fine, 12 was all that passed. Default settings in Macro 1 were more stringent than noted above.
|
|
|
bizgozcd
Junior Member
Posts: 53
Registered: 27-5-2020
Member Is Offline
|
|
Hello again. I have compiled a list of things I'm not totally clear on. Hoping for some help from the experienced users/Admin.
1. I do not fully understand the difference between Optimizing Price Data and TPD and Verification Price Data? Trying to understand the role each has
in the process so I can understand what happens if I use multiple options (timeframes and/or related products) in any/all the data streams above.
2. Nth Day Mode: What exactly does Trade/No Trade/All do? Should I be running 3 separate optimizations with these options, with everything else the
same to compare the results of each?
3. Is it “best practices” to avoid trading a seemingly profitable system that has a low ‘%-F’ (% profitable) if everything else lines up?
|
|
|
RandyT
Member
Posts: 123
Registered: 5-12-2019
Location: Colorado, USA
Member Is Offline
|
|
| Quote: Originally posted by bizgozcd | Hello again. I have compiled a list of things I'm not totally clear on. Hoping for some help from the experienced users/Admin.
1. I do not fully understand the difference between Optimizing Price Data and TPD and Verification Price Data? Trying to understand the role each has
in the process so I can understand what happens if I use multiple options (timeframes and/or related products) in any/all the data streams above.
2. Nth Day Mode: What exactly does Trade/No Trade/All do? Should I be running 3 separate optimizations with these options, with everything else the
same to compare the results of each?
3. Is it “best practices” to avoid trading a seemingly profitable system that has a low ‘%-F’ (% profitable) if everything else lines up?
|
bizgozcd, welcome to the struggle.
1.
a. Verify Price data is typically the same price data you have used in Optimization Price Data, however it is randomized with some noise. You can
create this data by right clicking in the price data list and clone with random noise. This data is used to verify the optimization results with
random data.
b. WF Price Data gives you the ability to use a different data set to run the WF optimization on. Different from what the data used to model the
system. I've never used this as it uses the same data if you leave it blank.
c. WF TPD I cannot explain and have never used.
2. I will take a stab at explaining Nth mode. This is also pretty well documented although it takes a few passes to fully understand. Basically, this
allows you to define some interval of data that the optimization process will pass over that data. So rather than create a date or percentage of data
to define your In Sample and Out Of Sample periods, Nth mode uses a portion for In Sample (Trd), then runs against the next period for Out of Sample
(NoTrd). So if you use a Nth Day =1 and a Nth Day period of 80, every 80 days of data will be used to test In Sample results against unseen 80 days of
data. That probably still doesn't make complete sense. I would recommend studying Peter's example of how he is using this in some of the most recent
CL system development.
3. My only comment regarding the low %-F may be an issue with stoploss being too tight. It personally makes me pretty nervous to buy into a system
that is right less than 50% of the time.
Hope that helps
Thanks received (2):
+1 getty002 at 2020-07-18 20:28:35 +1 bizgozcd at 2020-06-24 05:40:59
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
Just to chip in...
c, if you WF for example 29, 30 and 31 min data, then you need to specify TPD for WF, in this case 30min..
Btw, just saw that WF efficiency is added as a metric to sort on in fav macro, should be interesting to test as a robustness metric.. nothing to do
with the above though 
|
|
|
| Pages:
1
..
36
37
38
39
40
..
98 |
![[*]](images/davis/default_icon.gif){kind=icon}
