| Pages:
1
..
85
86
87
88
89
..
98 |
heyligerb
Junior Member
Posts: 40
Registered: 22-5-2017
Member Is Offline
Mood: No Mood
|
|
Recent Performance June 2022 - March 2023
Just curious if anyone out there has been able to build systems in US stock indices using GSB which have delivered out-of-sample performance in the
past few months. It's been a challenging market, but I've put together a group of systems that seems to have no lost money, but hasn't really made any
either.
I'm just getting back into the swing of GSB (last used it in 2017). And just ordered a new computer to make better use of it. Looking forward to
getting some real work done!
See attached...
|
|
|
ChuckNZ
Junior Member
Posts: 50
Registered: 22-12-2022
Member Is Offline
|
|
Not able to match trades
When doing some beta testing, I only saw strategies that had perfect matches between gsb and tradestion trade listings. But something has happened
on my system.
I'm running GSB 1.0.65.56 and TS 10.0. I'm getting some matches, some trades in GSB not in TS and some trades in TS not in GSB.
I just made the data files in TS and copied them over.... no help. I've run out of ideas and could really use some help.
I think it's only on my end, but perhaps you might want to run some match checks on your end.
Thanks for any suggestions.
|
|
|
ChuckNZ
Junior Member
Posts: 50
Registered: 22-12-2022
Member Is Offline
|
|
just for completeness.... i have tried standalone and manager/worker combinations. i have tried 4 different futures contracts. i have not tried
an older version.... will do that now.
rquote=9051&tid=229&author=ChuckNZ]When doing some beta testing, I only saw strategies that had perfect matches between gsb and tradestion trade
listings. But something has happened on my system.
I'm running GSB 1.0.65.56 and TS 10.0. I'm getting some matches, some trades in GSB not in TS and some trades in TS not in GSB.
I just made the data files in TS and copied them over.... no help. I've run out of ideas and could really use some help.
I think it's only on my end, but perhaps you might want to run some match checks on your end.
Thanks for any suggestions.[/rquote]
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
Quote: Originally posted by admin  | Hi Daniel
Im not aware of any such issues. If ts has multiple stops running, i believe the tightest applies. I assume mc is the same
If you have this issue, i see it as an execution error, not a easy language error
I believe the stops are all in the docs, and I have little faith in them helping in any way on a day trading system. Long term system much more likely
to work
how to use them is also in the docs or videos, cant remember as its years ago.
if you optimize the exit levels on an existing system you will see
many stop methods do nothing, others give much less dd at great expense to NP, some might help but affect very few trades |
Hi Peter, yes its true that tightest applies, but that a less good way in terms of risk management of the trade, 2 contracts to sell, and one
contracts in position, if all goes well, yet tightest applies.
Anyway, this is not the main point.
Main point is, we need an explanation of what each exit type actually does with human words.
Since the name itself does not lend the user to understand what it is or what it does.
There is no guide for this that makes it possible to read or understand as of this moment.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
@chuck
see this, but use \gsb\getdiff version 1.07 as its a match check tool designed for gsb instead of windiff
it consider 2 vs 1.9999999999 an exact match, while a text compare will not.
If your stuck (this is very advanced, but your are smart guy) I can help you monday
indicently, one of the big reasons for missmatch is ts gives -1 before the first 500 bars on things like closeD
If the first bars are wrong, everything later is wrong on some indicators.
Work is being done every day on improving this, and I will have updated list next week of whats still to be done.
@heyligerb
you should get a tiny improving using @nq 800 to 1500 instead of @nq.d 830 to 1500.
Been a tought year for day trading ES, but nq seems to be holding up better. Your results are reasonable considering how hard the market has been
recently.
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
| Quote: Originally posted by ChuckNZ | just for completeness.... i have tried standalone and manager/worker combinations. i have tried 4 different futures contracts. i have not tried
an older version.... will do that now.
rquote=9051&tid=229&author=ChuckNZ]When doing some beta testing, I only saw strategies that had perfect matches between gsb and tradestion trade
listings. But something has happened on my system.
I'm running GSB 1.0.65.56 and TS 10.0. I'm getting some matches, some trades in GSB not in TS and some trades in TS not in GSB.
I just made the data files in TS and copied them over.... no help. I've run out of ideas and could really use some help.
I think it's only on my end, but perhaps you might want to run some match checks on your end.
Thanks for any suggestions.[/rquote] |
FYI, there is still a very large amount of indicators in GSB that does not match, so I assume this is the reason, and I would assume is expected until
GSB gets all the indicators fixed.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
@daniel
I will update this
https://trademaid.info/gsbhelp/Exitmodes.html
Its simple
IflatRevAANDSFFlatRev = exit if Indicator flat (IE for long gone from positive to zero or negative) and SF gone from positive to zero or reverse)
Thanks received (1):
+1 BlackBox at 2023-04-22 20:22:51
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
Thanks, that is kind of you.
We have about 18 of them as of now i think, and many is just a letter combination, so "simple" is a strong word perhaps 
Thanks received (1):
+1 BlackBox at 2023-04-22 20:22:25
|
|
|
brk
Junior Member
Posts: 38
Registered: 19-3-2023
Member Is Offline
|
|
Running multiple systems on the same ticker
So I was wondering what happens when I have multiple systems running the same ticker and one goes long and the other goes short. Does TS take care of
those and keep them separate or close and shut them down? To complicate this what would happen if you had manual override and got out of one of the
trades? Would it still manage the other strategy?
I was considering running multiple ES and NQ strategies that aren’t negatively correlated but I have seen them generate opposite trades.
|
|
|
REMO755
Member
Posts: 181
Registered: 11-4-2021
Member Is Offline
|
|
| Quote: Originally posted by Daniel UK1 | | Quote: Originally posted by admin | Hi Daniel
Im not aware of any such issues. If ts has multiple stops running, i believe the tightest applies. I assume mc is the same
If you have this issue, i see it as an execution error, not a easy language error
I believe the stops are all in the docs, and I have little faith in them helping in any way on a day trading system. Long term system much more likely
to work
how to use them is also in the docs or videos, cant remember as its years ago.
if you optimize the exit levels on an existing system you will see
many stop methods do nothing, others give much less dd at great expense to NP, some might help but affect very few trades |
Hi Peter, yes its true that tightest applies, but that a less good way in terms of risk management of the trade, 2 contracts to sell, and one
contracts in position, if all goes well, yet tightest applies.
Anyway, this is not the main point.
Main point is, we need an explanation of what each exit type actually does with human words.
Since the name itself does not lend the user to understand what it is or what it does.
There is no guide for this that makes it possible to read or understand as of this moment. |
@Daniel UK1
Hello,
Please, what are the flags that do not match?
These flags must be removed.
Does it seem foolhardy to swap systems where the flags don't match? Are we then playing the lottery?
What is the point of using buggy indicators?
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
| Quote: Originally posted by REMO755 | | Quote: Originally posted by Daniel UK1 | | Quote: Originally posted by admin | Hi Daniel
Im not aware of any such issues. If ts has multiple stops running, i believe the tightest applies. I assume mc is the same
If you have this issue, i see it as an execution error, not a easy language error
I believe the stops are all in the docs, and I have little faith in them helping in any way on a day trading system. Long term system much more likely
to work
how to use them is also in the docs or videos, cant remember as its years ago.
if you optimize the exit levels on an existing system you will see
many stop methods do nothing, others give much less dd at great expense to NP, some might help but affect very few trades |
Hi Peter, yes its true that tightest applies, but that a less good way in terms of risk management of the trade, 2 contracts to sell, and one
contracts in position, if all goes well, yet tightest applies.
Anyway, this is not the main point.
Main point is, we need an explanation of what each exit type actually does with human words.
Since the name itself does not lend the user to understand what it is or what it does.
There is no guide for this that makes it possible to read or understand as of this moment. |
@Daniel UK1
Hello,
Please, what are the flags that do not match?
These flags must be removed.
Does it seem foolhardy to swap systems where the flags don't match? Are we then playing the lottery?
What is the point of using buggy indicators? |
Hi,
What you mean with "flags" you mean indicators right ?
There is absolutly no point in using indicators that does not Match, I know@Peter is working extremely hard to fix all indicators in GSB that does not
match.
I know @Nick has a list of all indicators that has errors and does not match..
Perhaps its a good idea @Nick to share that list if you have it available?
|
|
|
REMO755
Member
Posts: 181
Registered: 11-4-2021
Member Is Offline
|
|
Hello,
Please Peter, I ask you to go to a version of correct indicators.
I don't understand how knowing that there are indicators with errors is allowed to use this?
How hard is it to have a correct version with the correct indicators?
|
|
|
REMO755
Member
Posts: 181
Registered: 11-4-2021
Member Is Offline
|
|
| Quote: Originally posted by Daniel UK1 | | Quote: Originally posted by REMO755 | | Quote: Originally posted by Daniel UK1 | | Quote: Originally posted by admin | Hi Daniel
Im not aware of any such issues. If ts has multiple stops running, i believe the tightest applies. I assume mc is the same
If you have this issue, i see it as an execution error, not a easy language error
I believe the stops are all in the docs, and I have little faith in them helping in any way on a day trading system. Long term system much more likely
to work
how to use them is also in the docs or videos, cant remember as its years ago.
if you optimize the exit levels on an existing system you will see
many stop methods do nothing, others give much less dd at great expense to NP, some might help but affect very few trades |
Hi Peter, yes its true that tightest applies, but that a less good way in terms of risk management of the trade, 2 contracts to sell, and one
contracts in position, if all goes well, yet tightest applies.
Anyway, this is not the main point.
Main point is, we need an explanation of what each exit type actually does with human words.
Since the name itself does not lend the user to understand what it is or what it does.
There is no guide for this that makes it possible to read or understand as of this moment. |
@Daniel UK1
Hello,
Please, what are the flags that do not match?
These flags must be removed.
Does it seem foolhardy to swap systems where the flags don't match? Are we then playing the lottery?
What is the point of using buggy indicators? |
Hi,
What you mean with "flags" you mean indicators right ?
There is absolutly no point in using indicators that does not Match, I know@Peter is working extremely hard to fix all indicators in GSB that does not
match.
I know @Nick has a list of all indicators that has errors and does not match..
Perhaps its a good idea @Nick to share that list if you have it available? |
Daniel UK1
Do you trade real with indicators with error?
I think I'm missing something.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
@remo
its very hard to get indicators to match. Many years ago pre gsb most apps that made ts code were terrible match's
The reason is , its just so complex to get right. Most platforms like ts mc and nt have various bugs that have to be coded around.
the most common Nt data feed is lacking up/down volume and some indicators require this.
You cant get a perfect match in this case.
In recent years CL and a few others for the first time ever had negative prices.
For example TS trix when price is < zero gives divide by zero. All results after that is wrong.
CloseD, HIghD , Lowd functions before the first 500 bars gives -1, and if the first price is wrong, all other prices are wrong for some inidicators,
or they are wrong for some time and then match after time
Often there is rounding issues.
ie 0.12345678 vs 0.12345679 Thats due to numbers being stored in binary format.
The rounding errors can give from near perfect to significant errors.
Othertimes match check was done on the first 10,000 bars and its perfect. After more years a missmatch occurrs.
Roughly 2 months programing time has gone into this of late, and progress is now very rapid as one fix can fix multiple indicators, or the same
concept applied fixes multiple indicators
So in summary you always need to check gsb matches your platform results. You want a > 99% match.
There is also a high amount of human error that causes match problems, or changes in data. GSB has extensive diagnostic tools to check these issues
too.
Once ts / mc is done, then we go to NT. That I hope will be fast as 1/2 the fixes were in GSB, not ts code.
thanks all for patience
|
|
|
REMO755
Member
Posts: 181
Registered: 11-4-2021
Member Is Offline
|
|
| Quote: Originally posted by admin | @remo
its very hard to get indicators to match. Many years ago pre gsb most apps that made ts code were terrible match's
The reason is , its just so complex to get right. Most platforms like ts mc and nt have various bugs that have to be coded around.
the most common Nt data feed is lacking up/down volume and some indicators require this.
You cant get a perfect match in this case.
In recent years CL and a few others for the first time ever had negative prices.
For example TS trix when price is < zero gives divide by zero. All results after that is wrong.
CloseD, HIghD , Lowd functions before the first 500 bars gives -1, and if the first price is wrong, all other prices are wrong for some inidicators,
or they are wrong for some time and then match after time
Often there is rounding issues.
ie 0.12345678 vs 0.12345679 Thats due to numbers being stored in binary format.
The rounding errors can give from near perfect to significant errors.
Othertimes match check was done on the first 10,000 bars and its perfect. After more years a missmatch occurrs.
Roughly 2 months programing time has gone into this of late, and progress is now very rapid as one fix can fix multiple indicators, or the same
concept applied fixes multiple indicators
So in summary you always need to check gsb matches your platform results. You want a > 99% match.
There is also a high amount of human error that causes match problems, or changes in data. GSB has extensive diagnostic tools to check these issues
too.
Once ts / mc is done, then we go to NT. That I hope will be fast as 1/2 the fixes were in GSB, not ts code.
thanks all for patience
|
OK, so we talked about the fact that the errors are minimal, this is reassuring.
And wouldn't it be better to publish those indicators with errors? @DanielUK1
says that list
Thank you.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
the errors were published in the last week or two. Likely here or in the beta forum. A lot more have been fixed in the last few days, so the list
needs updating....
but quite a number more are coming soon as the fixs done will also apply to another few more indicators.
Next build will also mean all users need to upgrade to keep use of the free cloud, as it will give missmatch if gsb old mangers talk to new workers
(there is a look up table that doesnt allow un-compatable workers to talk to each other)
there is also a number of new features, i have yet to tes. ie GA time of day and day of week, and disabling of all primary filters.
|
|
|
ChuckNZ
Junior Member
Posts: 50
Registered: 22-12-2022
Member Is Offline
|
|
Manager does not agree with workers
Others have written about this and I have read suggestions of things to check.
I am running manager and workers on 1.0.65.56.
The problem occurs with numerous instruments, timeframes and indicators.
The manager generates a few thousand strategies using local and cloud workers.
As I scroll through the strategies in the grid, the values change for about 1 out of 8 strategies. More trades, less trades, more drawdown, less
drawdown.
As I see it, when I scroll through the strategies, the manager is recalculating and coming up with different values than the workers originally did.
I have tried old optimisation files. I have tried with and without secondary and tertiary filters.
I also tried running on some prior versions, making sure to match the manager with the local workers.
I've spent two very long days on this and cannot find the problem. I will continue to try to isolate the problem but it is very frustrating.
Any thoughts would really be appreciated.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
@chuck,
we can do anydesk.com tomorrow. Thanks for all your efforts, but its time for me to look at it.
As a test, you can also copy your manager settings to the standalone.
|
|
|
REMO755
Member
Posts: 181
Registered: 11-4-2021
Member Is Offline
|
|
Hello,
Have you tried checking the Windows region format? An idea, I had this problem before.
|
|
|
NickW
Junior Member
Posts: 29
Registered: 12-2-2020
Member Is Offline
|
|
| Quote: Originally posted by admin | @chuck,
we can do anydesk.com tomorrow. Thanks for all your efforts, but its time for me to look at it.
As a test, you can also copy your manager settings to the standalone. |
@Chuck, it might also be a problem with your macro. Make sure after making settings changes to dates, etc, that there is at least a backtest being
run as a last step (before exporting stats etc).
Although i sometimes get this aswell, but extremely hard to reproduce on my side.
I try not to touch the UI while the run is going on until the macro is completely finished.
Nick
|
|
|
ChuckNZ
Junior Member
Posts: 50
Registered: 22-12-2022
Member Is Offline
|
|
We can do an anydesk session, if you like. I tried the standalone, but the standalone won't disagree with itself.
| Quote: Originally posted by admin | @chuck,
we can do anydesk.com tomorrow. Thanks for all your efforts, but its time for me to look at it.
As a test, you can also copy your manager settings to the standalone. |
|
|
|
ChuckNZ
Junior Member
Posts: 50
Registered: 22-12-2022
Member Is Offline
|
|
Thanks, NickW, but I'm not running a macro.... just creating strategies.
| Quote: Originally posted by NickW | | Quote: Originally posted by admin | @chuck,
we can do anydesk.com tomorrow. Thanks for all your efforts, but its time for me to look at it.
As a test, you can also copy your manager settings to the standalone. |
@Chuck, it might also be a problem with your macro. Make sure after making settings changes to dates, etc, that there is at least a backtest being
run as a last step (before exporting stats etc).
Although i sometimes get this aswell, but extremely hard to reproduce on my side.
I try not to touch the UI while the run is going on until the macro is completely finished.
Nick |
|
|
|
ChuckNZ
Junior Member
Posts: 50
Registered: 22-12-2022
Member Is Offline
|
|
Hi Peter.... I have to apologise.... the problem DOES show up when running the standalone version. Unbelievble. As I scroll down the strategies,
some of them change values (net profit is the easiest to watch).
| Quote: Originally posted by ChuckNZ | We can do an anydesk session, if you like. I tried the standalone, but the standalone won't disagree with itself.
| Quote: Originally posted by admin | @chuck,
we can do anydesk.com tomorrow. Thanks for all your efforts, but its time for me to look at it.
As a test, you can also copy your manager settings to the standalone. |
|
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
Chuck, this is a known bug, its happens from time to time, and has been reported by me first time 2020. Programmer and Peter has received all files
from me during debug sessions, I assume its on the list of the programmer to fix. So just to be clear, what you and others are reporting about metric
values that is changing, is not something new, and is not because you are doing anything wrong, or because your setup is wrong, and its not something
you can fix. I have spent days, weeks, and more weeks trying to debug, and problem happens on both standalone and manager.
Its not a game changer, but I think it needs to be fixed.
|
|
|
ChuckNZ
Junior Member
Posts: 50
Registered: 22-12-2022
Member Is Offline
|
|
Problem with changing metrics (cont'd)
I have devoted several more hours to this project and, perhaps, came up with something of interest.
If I remove these two indicators (CloseOverLowestLow and CloseLessLowestLow) the problem goes away.
Here's what I have done to prove this to be true:
1. Removed the two indicators
2. Run GSB and generate 600 strategies.
3. Highlight them all and send them to statistics A
4. Scroll through the strategies (3 times)
5. Highlight them all again and send them to statistics B
6. Statistics A and B agreed.
Please let me know if my test sounds valid. I will now run the same test on a couple thousand strategies.
|
|
|
| Pages:
1
..
85
86
87
88
89
..
98 |
![[*]](images/davis/default_icon.gif){kind=icon}
