| Pages:
1
..
37
38
39
40
41
..
98 |
bizgozcd
Junior Member
Posts: 53
Registered: 27-5-2020
Member Is Offline
|
|
Thanks Randy and Daniel for the answers. Yes, I do understand that GSB is verifying the result from the optimization, but I'm trying to understand
this on a higher level. For example, what exactly does the verification do that the optimization didn't or what is it verifying? ie. if the
verification "result" is better than the original...or worse...or passes verification or fails....what happens? Does Verification process the original
optimization result somehow or is verification merely a reporting function for me to manually compare? Or...is the verification process changing the
results somehow?
WRT Nth mode...I have read the documentation on this multiple times already and watched the methodology videos lots of times. I am pretty sure I
understand the methodology process in using it, just not what is happening behind the scenes. Let me rephrase my question on this. I understand Nth
mode, that it is alternating periods of trade and no trade. But I do not understand what it is doing if I select Trade/No Trade/All. If I'm selecting
Nth mode (ie 1 and 80) is it alternating periods anyway regardless of what i select? If I select ALL, does that mean it's not using Nth periods, just
one big period? Does All process both the Trade and No Trade periods at the same time, thus saving me from running Trade and No Trade optimizations
separately, so that I can view the in sample and out of sample data separately or must I manually run fresh optimizations separately with each setting
and compare afterwards?
>> 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.
Totally clear.  Thank you.
Thanks again. I figure I got this far, may as well understand it as well as I can before implementing anything.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Verification puts the same system on different data.
it could be the same data with random noise, different time frame or different symbol.
You would expect the average result to degrade.
The criteria for verification is here.
So you can have vs thast says say 4 out of 8 streams passed or vss which is the % degradation.
For ES this worked well (been so long I forgot what one was best, I know it was random noise was best - 4 x 5 ticks and 4 x 10 ticks)
but for CL i didnt get improvement
with nth, if you build nth no trade, you can then click trade and you see the alt set of data which is out of sample.
if you put it to all, you see the IS and OOS.
You can also build systems with nth trade, and put the post trade setting to no trade. SO then whatever is in the GUI is now OOS. This is how we do
indicator testing.
Thanks received (2):
+1 cotila1 at 2020-06-25 13:24:40 +1 bizgozcd at 2020-06-25 09:50:47
|
|
|
bizgozcd
Junior Member
Posts: 53
Registered: 27-5-2020
Member Is Offline
|
|
Greetings once again.
So, I’ve had a few internet issues halfway through an optimization. Is there a way to stop the process so that I can reboot my computer and then
continue where I left off?
The only thing I was able to find was the auto save function, but if that's the best way to recover, how can I move those systems back to the Unique
Systems tab to perform macros?
Thank you.
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
Quote: Originally posted by bizgozcd  | Greetings once again.
So, I’ve had a few internet issues halfway through an optimization. Is there a way to stop the process so that I can reboot my computer and then
continue where I left off?
The only thing I was able to find was the auto save function, but if that's the best way to recover, how can I move those systems back to the Unique
Systems tab to perform macros?
Thank you.
|
Hi There,
Probably best to just mark all systems, then right click and save systems, then reboot, and rght click again in a new manager and then load all saved
systems to same opt setting and continue your process.. good luck
As a side note, A a good tip is to make at least 4-5 or more builds of 25-50k systems on same opt setting your are testing and get the averages of
the stats, so you make sure to take a decision on not random data..
Thanks received (1):
+1 bizgozcd at 2020-06-29 09:12:45
|
|
|
Carl
Member
Posts: 342
Registered: 10-5-2017
Member Is Offline
Mood: No Mood
|
|
My thoughts:
after the pc restart, start gsb, check if the loaded gsboptset file is the one you were using, and then load the saved strategies that were built
using this gsboptset file.
|
|
|
bizgozcd
Junior Member
Posts: 53
Registered: 27-5-2020
Member Is Offline
|
|
Thanks for the answers. If GSB starts again, does it start from the very beginning? Will the results be different every run? (As each optimization is
collecting only a sample of the possible systems).
|
|
|
Carl
Member
Posts: 342
Registered: 10-5-2017
Member Is Offline
Mood: No Mood
|
|
Yes, also depends or the "seed" GSB uses.
Peter advises using 4 or 5 different GSB runs.
|
|
|
bizgozcd
Junior Member
Posts: 53
Registered: 27-5-2020
Member Is Offline
|
|
I’ve had a few internet issues the last few days, sometimes requiring my computer to reboot to get back to normal.
So my question is, If I’m halfway through an optimization, is there a way to reboot my computer and then continue where I left off in GSB? I have
since set up the Auto Save Performance filter, but I would like to move all of those systems to the unique systems Tab on the main Dashboard page. Is
there a way to either restore with all the saved data or have GSB pick up where it stopped?
Thanks again.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by bizgozcd | I’ve had a few internet issues the last few days, sometimes requiring my computer to reboot to get back to normal.
So my question is, If I’m halfway through an optimization, is there a way to reboot my computer and then continue where I left off in GSB? I have
since set up the Auto Save Performance filter, but I would like to move all of those systems to the unique systems Tab on the main Dashboard page. Is
there a way to either restore with all the saved data or have GSB pick up where it stopped?
Thanks again. |
auto save is a pain, takes too much disk space.
If you havnt got the 250 systems, i would start again.
But if you have the 250 systems, you could save them and load them if you crash.
A internet issue shouldnt cause your machine to need to reboot. But regardless UN-planed reboots occur to most of us on occasions. Windows also needs
a reboot from time to time to be stable for many users. A once a week reboot is important on my live trading pc for stability
also to load 50k systems can take a very long time.
Consider buying another server, or rent a server is going to help your situation too.
|
|
|
Transatlantic
Junior Member
Posts: 3
Registered: 12-7-2020
Member Is Offline
|
|
Problems Opening TS Workspace for Free Systems
Hi,
I followed the instruction on the YouTube video correctly to the best of my knowledge and did create the custom session 0830_1500. I have that
session working correctly on a new chart I added.
When I open the TS Workspace 'GsbEQ+NQ Amended', I get an error in TS stating the following:
1. Selected Custom Session Template has been removed. The regular session will be used.
2. You may not mix symbols with different delays in the same window.
I think #2 is the problem since it appears to be opening a delayed version of @NQ (@NQ.D) yet I have real-time permissions for @NQ. I guess the
issue is with Data2 for $NDX.X?
Note I can open $NDX.X on a separate chart but I notice for some reason it actually uses $NDX.X(D).
Any help much appreciated. Thanks.
Scott
|
|
|
DocBober55
Junior Member
Posts: 17
Registered: 18-6-2020
Member Is Offline
|
|
When I open $NDX.X on a chart there is no indication that data is delayed [i.e. no (D) after the symbol - $NDX.X is what shows]. Evidently you are
getting delayed data for $NDX.X. You probably could straighten out the data issue by calling TradeStation support. Sometimes, when I update my data
subscriptions on-line it does not work properly.
|
|
|
Carl
Member
Posts: 342
Registered: 10-5-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by Transatlantic | Hi,
I followed the instruction on the YouTube video correctly to the best of my knowledge and did create the custom session 0830_1500. I have that
session working correctly on a new chart I added.
When I open the TS Workspace 'GsbEQ+NQ Amended', I get an error in TS stating the following:
1. Selected Custom Session Template has been removed. The regular session will be used.
2. You may not mix symbols with different delays in the same window.
I think #2 is the problem since it appears to be opening a delayed version of @NQ (@NQ.D) yet I have real-time permissions for @NQ. I guess the
issue is with Data2 for $NDX.X?
Note I can open $NDX.X on a separate chart but I notice for some reason it actually uses $NDX.X(D).
Any help much appreciated. Thanks.
Scott |
Hi transatlantic,
Putting NQ and NDX.X in the same TS-chart requires local time, not exchange time.
So start a new chart, choose NQ, choose local time, session 830-1500, add data2 NDX.X.
|
|
|
Transatlantic
Junior Member
Posts: 3
Registered: 12-7-2020
Member Is Offline
|
|
| Quote: Originally posted by DocBober55 | | When I open $NDX.X on a chart there is no indication that data is delayed [i.e. no (D) after the symbol - $NDX.X is what shows]. Evidently you are
getting delayed data for $NDX.X. You probably could straighten out the data issue by calling TradeStation support. Sometimes, when I update my data
subscriptions on-line it does not work properly. |
Agree and will see if subscribing to real-time Nasdaq corrects this.
Bizarrely TS is giving me $SPX.X real-time by default but is not giving me $NDX.X real-time.
|
|
|
Transatlantic
Junior Member
Posts: 3
Registered: 12-7-2020
Member Is Offline
|
|
| Quote: Originally posted by Carl | | Quote: Originally posted by Transatlantic | H
Hi transatlantic,
Putting NQ and NDX.X in the same TS-chart requires local time, not exchange time.
So start a new chart, choose NQ, choose local time, session 830-1500, add data2 NDX.X. |
Hi Carl,
Thanks for your help but that does not work. I just tried it again. The issue seems to be that regardless of how $NDX.X is specified it is using
$NDX.X(D) which seems to cause problems.
Scott
|
|
|
|
Carl
Member
Posts: 342
Registered: 10-5-2017
Member Is Offline
Mood: No Mood
|
|
Hi Scott,
Then the issue is caused by the fact you don't have a Tradestation subscription to the indices like NDX.X.
I think it costs an extra 2 or 3 USD a month.
But you can test the strategy with NQ as data2 as well.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by DocBober55 | | When I open $NDX.X on a chart there is no indication that data is delayed [i.e. no (D) after the symbol - $NDX.X is what shows]. Evidently you are
getting delayed data for $NDX.X. You probably could straighten out the data issue by calling TradeStation support. Sometimes, when I update my data
subscriptions on-line it does not work properly. |
I wouldnt bother with NQ.
When NQ was made from the ES system, we were aware the system was not as robust as the ES system. This turned out to be the case.
|
|
|
bizgozcd
Junior Member
Posts: 53
Registered: 27-5-2020
Member Is Offline
|
|
Hello again.
Making excellent progress, moving up the learning curve.
I've now used the methodology several times and I've come up with a system I want to paper trade on TS.
Copied the script over to TS and applied to a chart. TS is not closing out the trades end of day as is built into the system. I believe the session
settings are correct because I've been able to match up trades between GSB and TS on other systems I've tried and I'm using the same settings. Is
there anything else I need to look at? (Exit on Close is set to GSB, not TS, but I don't think that should matter as the Manual says difference is
shorter code for EL?) One last question...is there a way to change a single parameter within GSB or is editing the script required?
Many thanks.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by bizgozcd | Hello again.
Making excellent progress, moving up the learning curve.
I've now used the methodology several times and I've come up with a system I want to paper trade on TS.
Copied the script over to TS and applied to a chart. TS is not closing out the trades end of day as is built into the system. I believe the session
settings are correct because I've been able to match up trades between GSB and TS on other systems I've tried and I'm using the same settings. Is
there anything else I need to look at? (Exit on Close is set to GSB, not TS, but I don't think that should matter as the Manual says difference is
shorter code for EL?) One last question...is there a way to change a single parameter within GSB or is editing the script required?
Many thanks.
|
Great you made progress.
If your not trading at a ts brokeage account,
add setexitonclose; any where in the code will fix it.
check your code times match
is if last bar is 1500, then if time >= 1500 then {exit trades}
This time might be different. If they are differnet, you need to check all times in the code.
Check your last bar of day = the same time of the data in GSB
|
|
|
getty002
Junior Member
Posts: 30
Registered: 10-7-2020
Member Is Offline
|
|
Thanks, I'm in the same spot and had the same issue. I'll add that to the code to exit at close.
|
|
|
getty002
Junior Member
Posts: 30
Registered: 10-7-2020
Member Is Offline
|
|
This is a question unrelated to GSB, but unfortunately it won't let me create a new topic as a newbie. BTW, glad to join your community!
My question is related to the topic of how to identify when a strategy has decayed sufficiently to consider making changes. Does anyone have any TS
indicators or quantitative calculations for statistically evaluating when to make a strategy change? I watched this great video by "Trade-Systech"
which shows a T-statistic method for evaluating performance that's outside the distribution.
https://www.youtube.com/watch?v=635u9rolv4I&t=611s
It looks like he's also a GSB user? If you know who he is, perhaps send me his handle so I could IM him. As a reference, another good read on
statistical analysis of trading system performance is here (a bit old, 2014): https://www.stat.berkeley.edu/~aldous/157/Papers/harvey.pdf
Thanks in advance!
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by getty002 | This is a question unrelated to GSB, but unfortunately it won't let me create a new topic as a newbie. BTW, glad to join your community!
My question is related to the topic of how to identify when a strategy has decayed sufficiently to consider making changes. Does anyone have any TS
indicators or quantitative calculations for statistically evaluating when to make a strategy change? I watched this great video by "Trade-Systech"
which shows a T-statistic method for evaluating performance that's outside the distribution.
https://www.youtube.com/watch?v=635u9rolv4I&t=611s
It looks like he's also a GSB user? If you know who he is, perhaps send me his handle so I could IM him. As a reference, another good read on
statistical analysis of trading system performance is here (a bit old, 2014): https://www.stat.berkeley.edu/~aldous/157/Papers/harvey.pdf
Thanks in advance!
|
I dont know who it is, and the ts code at the end of the video doesnt look like GSB
His web site is no longer.
I have a different approach, and im not saying its not got room to improve.
I use pa pro (my product) and if the system has not made profit it the last year or pa pro just thinks it doesnt improve np/dd of the portfolio, the
system gets dropped. I dont know anyone who claims to be good at restarting systems that fail.
I just rarely trade more than one contract per system, so I get high diversified with lots of systems
|
|
|
getty002
Junior Member
Posts: 30
Registered: 10-7-2020
Member Is Offline
|
|
| Quote: Originally posted by admin | | Quote: Originally posted by getty002 | This is a question unrelated to GSB, but unfortunately it won't let me create a new topic as a newbie. BTW, glad to join your community!
My question is related to the topic of how to identify when a strategy has decayed sufficiently to consider making changes. Does anyone have any TS
indicators or quantitative calculations for statistically evaluating when to make a strategy change? I watched this great video by "Trade-Systech"
which shows a T-statistic method for evaluating performance that's outside the distribution.
https://www.youtube.com/watch?v=635u9rolv4I&t=611s
It looks like he's also a GSB user? If you know who he is, perhaps send me his handle so I could IM him. As a reference, another good read on
statistical analysis of trading system performance is here (a bit old, 2014): https://www.stat.berkeley.edu/~aldous/157/Papers/harvey.pdf
Thanks in advance!
|
I dont know who it is, and the ts code at the end of the video doesnt look like GSB
His web site is no longer.
I have a different approach, and im not saying its not got room to improve.
I use pa pro (my product) and if the system has not made profit it the last year or pa pro just thinks it doesnt improve np/dd of the portfolio, the
system gets dropped. I dont know anyone who claims to be good at restarting systems that fail.
I just rarely trade more than one contract per system, so I get high diversified with lots of systems |
Right, I will be diving into PA Pro soon once I feel like I've mastered GSB. Regarding diversification through many strategies, in practice, what's
the most number of strategies you'd consider putting on a single instrument?
The reason an indicator approach is attractive (in addition to your method, not in exclusion) is that I dynamically see there's potential problem when
monitoring in Tradestation. If I have say 30 charts up, an indicator method would provide a very fast way of identifying a suspect strategy in TS to
review more closely. In fact, I'd probably set an alert to get an email, if it were a reliable metric.
Thanks for your help.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Hi Getty
In a perfect world you have many markets ,time frames. However it takes us a long time to master new markets.
I have quite a few markets, but lots of them are from GSB build 2 or more years ago. These dont benefit from what we have learnt in the last 6 months.
So its a constant cycle of re-applying what we learn from one market, onto another market.
Some things are unique to each market, but many things are in common too.
|
|
|
Carl
Member
Posts: 342
Registered: 10-5-2017
Member Is Offline
Mood: No Mood
|
|
| Quote: Originally posted by getty002 | This is a question unrelated to GSB, but unfortunately it won't let me create a new topic as a newbie. BTW, glad to join your community!
My question is related to the topic of how to identify when a strategy has decayed sufficiently to consider making changes. Does anyone have any TS
indicators or quantitative calculations for statistically evaluating when to make a strategy change? I watched this great video by "Trade-Systech"
which shows a T-statistic method for evaluating performance that's outside the distribution.
https://www.youtube.com/watch?v=635u9rolv4I&t=611s
It looks like he's also a GSB user? If you know who he is, perhaps send me his handle so I could IM him. As a reference, another good read on
statistical analysis of trading system performance is here (a bit old, 2014): https://www.stat.berkeley.edu/~aldous/157/Papers/harvey.pdf
Thanks in advance!
|
Hi getty002,
The T-statistic shows if the sample (i.e. most recent 30 trades) could be part of the population (i.e. all trades sofar without the most recent 30
trades) based on average and standard deviation.
If the sample is an outlier with a small chance of only x% that it could be part of the population, you could consider stop trading this strategy.
More info here, https://en.wikipedia.org/wiki/T-statistic
I know who the guy in the video is on the GSB forum.
He follows the forum every now and then.
So if he feels like it, he can react to these posts.
And an updated version of the 2014 document you mentioned can be found here:
https://papers.ssrn.com/sol3/papers.cfm?abstract_id=2474755
|
|
|
winbolton
Newbie
Posts: 1
Registered: 10-7-2020
Member Is Offline
|
|
gaAutoStopped
My systems are stopping prematurely. under optimization status on the right hand side, the status = gaAutoStopped
It seems to only happen when I change any of the stop losses from false to true
I cant find any documentation on it. What does this mean?
|
|
|
| Pages:
1
..
37
38
39
40
41
..
98 |
![[*]](images/davis/default_icon.gif){kind=icon}
