| Pages:
1
..
89
90
91
92
93
..
98 |
thowoc213
Junior Member
Posts: 39
Registered: 22-8-2021
Location: Poland
Member Is Offline
|
|
Hi,
Is the GSB cloud down?
I run a job today and GSB Manager couldn't find any cloud worker.
Thomas
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Hi Thomas
there are no issues. What build are you on? 65.31 or 65.56
|
|
|
thowoc213
Junior Member
Posts: 39
Registered: 22-8-2021
Location: Poland
Member Is Offline
|
|
I'm on 65.56
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
No issues that im aware off. I did use my own cloud a little today but there still should have been plenty. Anyone else got a comment of workers or no
workers?
Ive not done system building for a month nearly, so should be lots of workers free
|
|
|
thowoc213
Junior Member
Posts: 39
Registered: 22-8-2021
Location: Poland
Member Is Offline
|
|
I can now see that GSB cloud is up.
My computer runs 4 jobs for a GSB user.
So, the issue must be something odd with my Manager.
Which is strange because I could access the GSB cloud 12 hours back.
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
No problem here, been running on cloud and all ok here..
Thanks received (1):
+1 admin at 2023-06-27 23:04:29
|
|
|
thowoc213
Junior Member
Posts: 39
Registered: 22-8-2021
Location: Poland
Member Is Offline
|
|
@Peter
@Daniel
After computer reboot the problem was gone. I have access to the GSB cloud.
What the problem was, have no idea.
Thomas
|
|
|
thowoc213
Junior Member
Posts: 39
Registered: 22-8-2021
Location: Poland
Member Is Offline
|
|
Peter,
I am always in doubt when doing a WFA. The problem is how to set the size of the steps for the parameters.
For example, a system has 3 parameters to optimize.
The parameter values from the GSB process are: p1 = 12, p2 = 150, p3 = 80. The optimization windows have 50% of values on each side, which means they
are: p1(6-18), p2(75-225) and p3(40-120).
What step size to choose for the parameters? The choice for p1 should obviously be 1. What are the steps for p2 and p3 with much wider windows? Should
the step sizes for p2 and p3 be ‘normalized’ to ‘fit’ p1, or they might very well also have a step size 1?
By ‘normalized’ I mean that other parameters should have the same number of steps within their respective windows. Thus, since p1 has 10 steps within
its window, the ‘normalization’ would result in p2(75-225,15) and p3(40-120,8).
I assume the GSB has an algorithm for choosing correct steps for different ‘wideness’ of the window.
Could you please share how the GSB does that?
Thanks in advance.
Thomas
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
@thowoc213
I would probably use 200%
I cant remember the logic for the step size. Its been many years.
If you right click a system and do ai wf parameter analysis then you will see the steps used
I really like manual optimization of each input one at a time, but its very time consuming.
ai wf however makes it much faster
However manual optimization is for advanced users and im not keen to do all the support on that as its complex and requires skill and experience
the other issue is ga pop and generations
if you have entry types aic, there are no weights so 100 x 100 is plenty
if you did say 5 indicators and entry type cross, you would need much more. (im not aware of anyone doing this)
Thanks received (1):
+1 thowoc213 at 2023-06-28 09:49:31
|
|
|
TraderTom
Junior Member
Posts: 13
Registered: 25-6-2023
Member Is Offline
|
|
Success Stories
Hi all, I'm a potential purchaser of GSB, doing research before I commit. I wanted to post a new topic for this, but even though I'm registered on
the forum, it doesn't seem to allow me to do so, so I'll post here. Maybe Peter can move this elsewhere if this isn't the right place.
What I'd like to ask if there are some current GSB users that are willing to share their success stories. I've read the testimonials posted at https://www.trademaid.info/testimonials.html, but those seem a bit dated. And I was hoping for maybe having the opportunity to ask a couple of
follow up questions.
I'm not asking for any proprietary details, of course. What I'd like to know are things like how long have you been using GSB? How long did it take
you to produce your first profitable system? How long did that system stay profitable, or is it still profitable? How many different GSB systems are
you trading now? If you care to, can you indicate which markets you've found to be the easiest for GSB to develop systems for? And do you have any
other experience with genetic trading system builders, and if so, how does GSB compare? I notice there are a number of other add-on products for GSB
for sale, how essential are those? And finally, any words of advice for a potential new user?
Thanks very much in advance!
|
|
|
NickW
Junior Member
Posts: 29
Registered: 12-2-2020
Member Is Offline
|
|
Hi,
A picture is worth a thousand words. See screenshot of my oldest system in my database. Not the prettiest, but the first. It was developed Dec 2020
and not touch after that. Only changes I've made was changing stop loss as back then I used to use the biggest stop, not anymore. Green line is OOS
start date.
Equity markets easier. There's a good methodology on NQ in the documentation. GSB has a learning curve, but once you understand it all, then it's
great. But i'm trading Equities, Metals and Oil and Gas markets. I have about 300 systems in my arsenal to choose from across all these markets.
Tried everything under the sun until GSB. GSB is very unique and the only thing I found working.
Good luck
Nick
Quote: Originally posted by TraderTom  | Hi all, I'm a potential purchaser of GSB, doing research before I commit. I wanted to post a new topic for this, but even though I'm registered on
the forum, it doesn't seem to allow me to do so, so I'll post here. Maybe Peter can move this elsewhere if this isn't the right place.
What I'd like to ask if there are some current GSB users that are willing to share their success stories. I've read the testimonials posted at https://www.trademaid.info/testimonials.html, but those seem a bit dated. And I was hoping for maybe having the opportunity to ask a couple of
follow up questions.
I'm not asking for any proprietary details, of course. What I'd like to know are things like how long have you been using GSB? How long did it take
you to produce your first profitable system? How long did that system stay profitable, or is it still profitable? How many different GSB systems are
you trading now? If you care to, can you indicate which markets you've found to be the easiest for GSB to develop systems for? And do you have any
other experience with genetic trading system builders, and if so, how does GSB compare? I notice there are a number of other add-on products for GSB
for sale, how essential are those? And finally, any words of advice for a potential new user?
Thanks very much in advance!
|
|
|
|
TradingPrice
Junior Member
Posts: 21
Registered: 4-8-2022
Member Is Offline
|
|
@admin
is AI WF analysis explained anywhere in documentation? if not could you please tell us more?
it seems basically like Tradestation manual optimization of parameters? as in it gives you NP & AT if you change indic. parameters?
|
|
|
TraderTom
Junior Member
Posts: 13
Registered: 25-6-2023
Member Is Offline
|
|
Hi Nick,
Thanks much for the reply. That's exactly the sort of thing I'm looking for. I particularly like your approach of having a lot of systems to use.
That was the reason for my initial interest in GSB. I believe that "diversification is the Holy Grail of trading" (to quote Ray Dalio), but I was
struggling with how to develop lots of different systems that are not too correlated. It would take too long to do manually, so I'm looking for an
automated solution, which is where I'm thinking GSB could come in.
Great to hear of your success story, I hope there are more. Thanks.
P.s. Mind telling me what you're trading in the system for that screen shot? If not, no worries.
|
|
|
NickW
Junior Member
Posts: 29
Registered: 12-2-2020
Member Is Offline
|
|
That screenshot is a system trading ES.
Thanks received (1):
+1 TraderTom at 2023-06-28 17:28:09
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Hi TraderTom
Welcome
Lots of good questions, here is a few comments.
I locked new threads. We got so much duplication, total mess and no one gets an email on a new topic you create. Put a topic in a thread people
subscribe to and hopefully you get an answer.
I do not think of my self as a marketer, or software vendor. I am a trader who uses GSB for my trading, my cleints are paying for the programmers,
they get many but not all of there own suggestions added into the product, we share how we use GSB best
and its a circular loop thats in everyone's interest. Hence no constant emails of market, specials, discounts, updates on videos etc. Documentation
however tends to be fairly current as its easy to update
Somewhere a few months ago I published a few broker statements while in the middle of a nasty draw-down. Ive been trading new for 23 years
As for other products.
GSB automation saves massive amount of time and human error.
PA pro useful to combine systems, work out portfolios
EWFO really powerful for doing more through walk forwards if you have the time and skill to do them
GSB resource manger, free with gsb purchase or maintenance.
Thanks received (1):
+1 TraderTom at 2023-06-29 08:46:31
|
|
|
TraderTom
Junior Member
Posts: 13
Registered: 25-6-2023
Member Is Offline
|
|
Thanks for the reply, Peter.
It's starting to sink in that the GSB ecosystem is different from anything else out there, in terms of organization, incentives, and operation. I'm
slowly getting my head around it.
I've been looking at the costs associated with a GSB commitment. Here's my draft list, with notes. I'd appreciate any comments you or any users may
have.
Main GSB license: $1,500 + $200/year maintenance
Portfolio Analyst: $350 May not be needed initially, but seems useful to have later after a bunch of systems have been developed.
EWO: $390 I'm not clear on this, but I *think* this may not be needed for systems developed with GSB. Is that right?
GSB Automation: $700 This may not be needed initially, but it seems essential at some point down the road, probably sooner rather than later. But I'm
curious, how have other users done without this?
Trademaid Automation for TS: $500/year. I'm not really clear what this does, exactly. I do use TS as my broker, so it might be of interest at some
point.
New hardware: $2-3K I currently have an old Xeon E5-2609 machine with 32GB RAM, running Win 10. Is this enough to at least get started with GSB,
without a huge amount of frustration?
So, on the low end it looks like a $1,500 expense, for just GSB.
On the high end, it's $2,940 for software (everything except TM automation) and say $2,500 for hardware, so $5,440 total. Does this seem right?
Thanks to all for any feedback!
|
|
|
TradingPrice
Junior Member
Posts: 21
Registered: 4-8-2022
Member Is Offline
|
|
@TraderTom
- I bought GSB last Oct, so I dont consider myself an expert in GSB
- I don't have "a few years success story" like Nick, but so far so good with GSB, I like the software! trading only GSB systems and up for the year
so far (ES and NQ only) still many markets for me to research and build systems.
- GSB has a learning curve just like Nick said, it's definitely not plug and play (altho NQ methodology is solid and research had already been done
for everyone making it super easy to build systems for that market/methodology)
- if you haven't done trial, you should definitely try it. otherwise, all you need is GSB license to get started, other software can be bought later
if you like it
- for hardware, I'd get a 13th Gen Desktop i9 or i7. but you can still rent servers from @admin
- Portfolio Analyst is a must if you're a systematic trader, it's a separate software and you can add to it GSB systems + Non-GSB systems if you like
- GSB Automation is great to research markets (@admin does a community research once in a while and shares results with people who contribute)
- Trademaid Automation for TS is great if you're trading with Tradestation but not necessary, what I like most about it is symbol substitution. e.g.
trade MNQ from a chart of @NQ
- Cant speak of EWO as I haven't' tried it but I think it's also separate software meaning works with GSB systems + Non-GSB systems (GSB can do Walk
Forward on its own)
Best of luck
Thanks received (1):
+1 TraderTom at 2023-06-29 14:46:54
|
|
|
thowoc213
Junior Member
Posts: 39
Registered: 22-8-2021
Location: Poland
Member Is Offline
|
|
Hi Peter,
I'd appreciate clarifications regarding the “Post build periods” feature. The link to the GSB Guide is here: https://trademaid.info/gsbhelp/Posttradingperiods.html.
I don’t know how to insert pictures into a post here so my questions will be difficult to follow without seeing the GSB Guide page. Sorry for that.
The text from the Guide is between ##GSB Guide ##, my questions are between >>my questions<<.
## Auto date mode will optional turn date or nth day mode to true in the manager when systems are built, even though it was false in the blue
section(s) of trading periods.worker .##
>>Does the optionally mean that nothing happens if the fields are left blank? Why do you refer to both the Manager and the Worker? Since a worker does
a job on Manager behalf, it cannot have different settings than the Manager. Or? <<
## If nth was no trade in blue, in manager, then the system will be build as build nth no trade, but will appear in the manager using nth trade. ##
>> So, the Manager asks a worker to build systems using every second 80 days period. What the Auto Nth Dat=Trd will do? Sorry, I don’t understand
this at all. <<
## This means all the metrics you see in the manager are 100% out of sample. ##
>> How can any system be 100% out of sample? <<
## So if nth was all trade in blue, in manager, then the system will be build as build nth trade no trade, but will appear in the manager using nth
ALL. ##
>> The system will be build using every second 80 days period but will appear using all? And why this pointing out that it will appear in manager.
Shouldn’t we always look in the Manager? Sorry, I’m confused. <<
## However when the system is rebuilt in the manager, it will be calculated as nth being all. This can save a lot of time in the build process as to
change say 50,000 systems from nth no trade to trade takes some time. ##
>> So, we have 50k systems and we want to do some postprocessing. Say 300 are moved to Fav A and we want to see the whole period. It’s done by Ctr-A,
right mouse button down and select Dates=All from the pop-up menu. What good does the setting in red do? Is it for use in macros? Please explain. <<
By the way, BIG,BIG THANKS for the AI WF, it’s really a great function.
I’ve spent hours optimizing parameters one by one in TS and then analyzing the results in Excel. Now I’m done in minutes!
Thomas
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Hi Thomas
to add a picture, do preview post then upload file
Ive tried to reply as best I can but without arrows to specific fields in picitures its not easy.
The docs are correct about 100% out of sample. 80 days system was built on, but then nth is flipped and 100% pf this data is unseen
if nth post build was set to all, then 50% is out of sample (pre build nth 80)
in the work the setting could be nth no trade, and so fitness is derived from every 80 days of data, 80 days unseen
but if nth post build is all in the manger, the manger will see all periods and fitness derived from all days though half this data is unseen
"## So if nth was all trade in blue, in manager, then the system will be build as build nth trade no trade, but will appear in the manager using nth
ALL. ##
>> The system will be build using every second 80 days period but will appear using all? And why this pointing out that it will appear in manager.
Shouldn’t we always look in the Manager? Sorry, I’m confused. <<"
im pointing out how the process works so you understand the deeper workings. SO while you always look in the manager, this is explaining whats under
the lid.
Its important that you have out of sample periods, and no 'leak's from in sample to out of sample.
Its excellent you try to understand this, but worse case you could treat it as a black box and just follow the methadolgy
Glad you like AI paramater analysis
|
|
|
thowoc213
Junior Member
Posts: 39
Registered: 22-8-2021
Location: Poland
Member Is Offline
|
|
| Quote: Originally posted by admin | Hi Thomas
to add a picture, do preview post then upload file
Ive tried to reply as best I can but without arrows to specific fields in picitures its not easy.
The docs are correct about 100% out of sample. 80 days system was built on, but then nth is flipped and 100% pf this data is unseen
if nth post build was set to all, then 50% is out of sample (pre build nth 80)
in the work the setting could be nth no trade, and so fitness is derived from every 80 days of data, 80 days unseen
but if nth post build is all in the manger, the manger will see all periods and fitness derived from all days though half this data is unseen
"## So if nth was all trade in blue, in manager, then the system will be build as build nth trade no trade, but will appear in the manager using nth
ALL. ##
>> The system will be build using every second 80 days period but will appear using all? And why this pointing out that it will appear in manager.
Shouldn’t we always look in the Manager? Sorry, I’m confused. <<"
im pointing out how the process works so you understand the deeper workings. SO while you always look in the manager, this is explaining whats under
the lid.
Its important that you have out of sample periods, and no 'leak's from in sample to out of sample.
Its excellent you try to understand this, but worse case you could treat it as a black box and just follow the methadolgy
Glad you like AI paramater analysis
|
Thanks for the answer.
Well, the “black box” is good, but understanding is better. 
I’m experimenting with different settings just to get a deep understanding of GSB. What triggered my question about ‘Post Build’ was an observation I
made recently.
I took a copy of the Trade List, pasted it into Excel. In Excel I wrote a simple VBA routine checking the difference between trading dates from the
start (in one column) and difference between two consecutive trades (in another column).
I expected to see two things:
1. the jumps on 80 days or more where NoTrd period were started.
2. the gaps (80 days or more) between two consecutive trades, also NoTrd periods occurred.
I’ve found none of them. The list was without jumps and gaps. Just to check that I’m not doing something wrong, I took another script with many trades
(more that 2000) and checked it.
The result was the same, no jumps and no gaps.
Both scripts were taken from ‘Unique Systems’, which means that whatever I had in the ‘Post Build’ should be of no consequence.
There is a possibility that I unintentionally changed the Nth settings and things went wrong. Anyways, I will follow the ‘black box’ and do the same
check again.
Now the question about AI WF.
There are also checks of weights. I never thought of optimizing them, as they always have the same values. Now I see from the AI report that there are
indeed differences if weights are optimized. I made a test on a script that has excellent stability values (over 90) from WFO and when I took weights
into WF optimizations, the results were much worse. What is right thing top do, leave the weight out or have them optimized too?
Thanks!
Thomas
|
|
|
TraderTom
Junior Member
Posts: 13
Registered: 25-6-2023
Member Is Offline
|
|
Hi TradingPrice,
Thanks for the overview of the software, and how things fit together. I'm feeling more confident that I can just start with GSB and add more pieces
later, if/when needed, which is what you said.
My plan is to definitely do the 2-week GSB trial first. I have vacations coming up for the first half of July, so it will probably be late July
before I have an uninterrupted block of time to dig into this. In the meantime I will continue to monitor the forum, and hopefully a couple of more
users can post their experiences with GSB. There's nothing like having an actual live trader tell you it works to give you confidence!
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
@2thowoc213
if there are no gaps, then the post build settings must be set to all.
if you after systems are build, change the systems from nth trade to no trade, you will see the in sample have significantly higher metrics than the
out of sample.
If you are on nasdaq or ES, weights should not be used. Entry type is anyindicatorcrossed, and this does not use weights
cross entry type will work, but then weights are used. Cross is used for Gold (any many other markets). I normally optimize weights when used if aic
is not used.
|
|
|
thowoc213
Junior Member
Posts: 39
Registered: 22-8-2021
Location: Poland
Member Is Offline
|
|
| Quote: Originally posted by admin | @2thowoc213
if there are no gaps, then the post build settings must be set to all.
if you after systems are build, change the systems from nth trade to no trade, you will see the in sample have significantly higher metrics than the
out of sample.
If you are on nasdaq or ES, weights should not be used. Entry type is anyindicatorcrossed, and this does not use weights
cross entry type will work, but then weights are used. Cross is used for Gold (any many other markets). I normally optimize weights when used if aic
is not used.
|
Hi Peter,
Yes, post build settings were set to ALL.
I did plenty of tests with different setting combinations and I think I finally got it all.
Thanks for explaining how to use weights but there is one more question.
What do you mean by saying "I normally optimize weights when used if aic is not used"
Thomas
|
|
|
thowoc213
Junior Member
Posts: 39
Registered: 22-8-2021
Location: Poland
Member Is Offline
|
|
| Quote: Originally posted by TraderTom | Hi TradingPrice,
Thanks for the overview of the software, and how things fit together. I'm feeling more confident that I can just start with GSB and add more pieces
later, if/when needed, which is what you said.
My plan is to definitely do the 2-week GSB trial first. I have vacations coming up for the first half of July, so it will probably be late July
before I have an uninterrupted block of time to dig into this. In the meantime I will continue to monitor the forum, and hopefully a couple of more
users can post their experiences with GSB. There's nothing like having an actual live trader tell you it works to give you confidence!
|
The strategies from GSB served me well.
The last 3 months of 2021 gave nice profit, the whole 2022 was very good. This year though, I’m still in a 25% DD that now seems to recover. I trade
ES, NQ, GC, and CL.
The learning curve is steep, but you can just follow Peter’s methodology and learn GSB that way. It will be a little easier.
Don't hope too much for the 2 weeks trial, it's too less time.
I also use EWFO and Portfolio Analyst Pro. I recommend you to purchase that package.
Good luck!
Thomas
|
|
|
TraderTom
Junior Member
Posts: 13
Registered: 25-6-2023
Member Is Offline
|
|
Hi Thomas,
Thanks for the reply, I appreciate the information very much.
However, there is some information I came across recently that has left me quite concerned about GSB, and which I hope Peter can address. I was on
the Striker web site, and I saw they are tracking 12 different GSB systems, which you can rent to trade with them if you should want to. What
bothered me is that out of those 12 systems, only 5 were profitable over the tracking period, and the other 7 lost money. Of the 5 winners, there was
one superstar (GSBsys1ES), but none of the other 4 had made over $10K, trading 1 car. That was disappointing, both in how few winning systems there
were, and how relatively small the winnings are.
I hope I'm missing something here, and that I can expect better performance than that. From where I sit, to get going with GSB I'm looking at a
substantial investment in software and hardware, and a steep learning curve, so the rewards have to worth the price. Right now, it seems to me they
fall short. But I hope I'm wrong!
|
|
|
| Pages:
1
..
89
90
91
92
93
..
98 |
![[*]](images/davis/default_icon.gif){kind=icon}
