admin
Super Administrator
Posts: 5069
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
GSB first impressions
Please put your comments on how you find GSB, and the version number so users can see the direction of feedback
|
|
|
cotila1
Junior Member
Posts: 78
Registered: 8-5-2017
Member Is Offline
Mood: No Mood
|
|
''I'm pretty confident on great growing of GSB. In this early stage of GSB I have done a severe test on about 15 strategies on @ES.D (30 min T.Frame)
created with GSB: from original input file to GSB I have removed the last 2 years of data (just deleted from txt file). Meaning that GSB have worked
on known data till end of march 2015. In other words I have created a new un-seen period from apr-2015 till may-2017. On 100% of strategies, once the
code has been reported in TradeStation, they have continued to make profit on these 2 last years. Which is a great a result: 100% of strategies have
worked on 2 years of new OoS.
Those Strategies has been built with zero (or almost zero) cost of slipagge and commissions. At this early stage the inclusion of realistic costs is
not very much in scope yet, but I'm confident that the work in progress of GSB will include more filters to reduce the number of trades in order to
consider also those costs. But again its a super start for GSB. Congrats Peter, please continue!!!!
|
|
|
Carl
Member
Posts: 343
Registered: 10-5-2017
Member Is Offline
Mood: No Mood
|
|
Hello Peter,
I am really impressed with the results GSB is producing.
As mentioned before, the way you/GSB are using indicators is completely new to me, but seems to work much better than "the default way".
How did you come up with this method?
With version 20.1 I am a bit more concerned about the number of parameters used in a generated systems.
You were able to lift a simple GSB entry with a low average trade, to acceptable/good levels by using filters you call "truisms".
Why did you choose adding more oscillators in version 20.1 instead of truisms?
Thanks,
Carl
|
|
|
admin
Super Administrator
Posts: 5069
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Hi Carl,
thanks for your comments.
"How did you come up with this method?"
I have been building GSB for many years, and this generation is a total re write from scratch.
The earlier GSB worked well, but was dreadful to use. All command line driven, no gui, big ini file, no walk forward.
Earlier GSB was built with my knowledge of many similar products I had seen (both retail and private system builders), and also improved significantly
by a Russian mathematician.
There is still lots of work on the architecture to go, and current architecture may be one of several to come, or hybrids.
"With version 20.1 I am a bit more concerned about the number of parameters used in a generated systems."
20.1 has up to 5 indicators (from 3) and the addition of a secondary filter.
RC2ES has been on futures Truth top 10 table for the last 6 months. It has 5 indicators + a higher than normal amount of secondary filters.
So this proves complex systems can work well out of sample, though simple is always a preference.
The methodology & architecture coupled with insistence on doing walk forward lend itself to having much higher out of sample success than other system
builders.
Part of the methodology is to check that any indicator is not redundant or makes no improvement to the overall system. Redundant & invalid logic is a
significant reason for system builders failing out of sample, and having code that is way to complex.
You were able to lift a simple GSB entry with a low average trade, to acceptable/good levels by using filters you call
"truisms".
The gsbsys1 (see GSB Tutorial_number2a_gsb1_es30.xlsx) raised profit per trade of a 3 indicator system from $20.88 to $232.02 with adding 1 truism,
and optimizing the parameters.
Why did you choose adding more oscillators in version 20.1 instead of truisms?
In going through my notes on the old GSB, i noted that the best results were with about 5 indicators.
Typical profit factor and average trader were higher with the addition of more indicators.
Over 5 is to complex to walk forward.
I want to add the truisms too and considered more indicators and truism both essential.
|
|
|
moveo
Josef
Posts: 20
Registered: 10-5-2017
Member Is Offline
Mood: No Mood
|
|
Peter,
when I asked a year ago for a payable systembuilder
I had no idea how beneficial the GSB would be.
Now, testing GSB for a couple of weeks I am
really surprised how great this works.
Systems developed with GSB are working
outstanding robust on unseen data.
Sure, it is still tricky to pick a system that performs
good in the future, but with this great piece of
software and with ewfo it becomes more doable.
Well done and thank you for sharing your work
and your software with retail and private traders
Moveo
|
|
|
Gregorian
Junior Member
Posts: 97
Registered: 23-5-2017
Member Is Offline
Mood: No Mood
|
|
Very impressed so far with GSB. I've accomplished more in the way of developing promising systems with GSB in two weeks than I have been able to do
after months of working with Adaptrade and years with StrategyQuant. It compares very favorably with TSL and in some ways exceeds TSL's ability to
produce robust OOS performance.
I especially appreciate your quick attention to enhancements and bug fixes. I am grateful too that you use “on close” orders; that’s so
essential for the Advanced Bar Types, yet Adaptrade and StrategyQuant don’t seem to understand how important that is. Keep up the good work!
|
|
|
Carl
Member
Posts: 343
Registered: 10-5-2017
Member Is Offline
Mood: No Mood
|
|
Hi Gregorian,
Good to hear you were able to develop robust strategies with GSB.
GSB has developed a bunch of good systems on ES for me too.
On what tickers did you manage to get robust systems?
Thanks
|
|
|
Gregorian
Junior Member
Posts: 97
Registered: 23-5-2017
Member Is Offline
Mood: No Mood
|
|
Carl - Best results so far have been with CL with Point bars on MC, but please don't read too much into that. It's still very early on in the testing
to draw conclusions. I would rather trade ES but have not had success on any Advanced Bar Types with it yet.
|
|
|
Carl
Member
Posts: 343
Registered: 10-5-2017
Member Is Offline
Mood: No Mood
|
|
Hi Gregorian,
Sofar I have only used default bars for ES in GSB.
GSB produced several strats with 80k-100k net profit over 20 years and a stoploss of 350 USD.
Did you purchase GSB? If so, what do you think about the performance of Peter's GSBsys2ES code he shared?
Or do you only trade advanced bar types?
Carl
|
|
|
rws
Member
Posts: 114
Registered: 12-6-2017
Member Is Offline
Mood: No Mood
|
|
Hi Gregorian,
Do you own TSL? How long did you use it?
Thanks
|
|
|
Gregorian
Junior Member
Posts: 97
Registered: 23-5-2017
Member Is Offline
Mood: No Mood
|
|
Carl - Yes, I purchased GSB. I only trade Advanced Bar Types and am only interested in strategies developed to meet my specific criteria.
rws - I have been using TSL since March, 2015. Mike provides excellent support, as he should for the price charged, but the product has not been
advancing in ways that benefit me, plus its OOS performance is not so good. However I am a much more informed user of TSL, Adaptrade, StrategyQuant,
and now GSB as a result of what I've learned from Mike. To make use of these tools properly, one must have a fairly thorough understanding of order
types, price feeds, and role of spreads, all of which has taken this slow learner some time to grasp.
|
|
|
admin
Super Administrator
Posts: 5069
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Hi Gregorian, You sound like you have a great deal of depth with system builders, so it will be interesting to see how you progress with GSB. Its such
a powerful tool, how its used is a universe to explore. Especially in light of all the things that arnt yet done.
Secondary filters / Truisms and two pass WF with auto detect fitness are things Im keen to see.
Short term focus is bug fixes & small tweaks, then secondary filters.
There also seems to be significant interest in 3rd party / propriety indicators. Most likely we will allow
GSB to use propriety indicators with lengths / values that GSB can genetically optimize. This is much smarter than csv export of fixed indicator
values into GSB
with no options to shift parameters. This will take an entire weeks programing work, so is a few months away.
|
|
|
kelsotrader
Junior Member
Posts: 29
Registered: 16-2-2018
Location: Tapanui - New Zealand
Member Is Offline
Mood: No Mood
|
|
GSB Feedback 1.0.43.40
As you have asked for some feedback and I know how important it is I would like to make the following comments.
I have been involved in computers for over 35 years both in hardware repair and software development but I am new to both TradeStation and GSB so my
comments should be taken with this in mind.
I am a visual learner so I really appreciated the excellent videos that you have produced.
As I am more or less retired I was able to devote a lot of time in watching videos, reading the forums and what documentation there was and in
duplicating creating a strategy and running it on TradeStation.
I did however find it a frustrating exercise at times.
First off for a person starting out with GSB there are dozens of acronyms used throughout both GSB and the forums without an acronyms reference guide.
So a list of both acronyms and their usage, importance etc, would be a great help.
I know that you intend to update the documentation is time but I think that it needs to be done sooner rather that later. (Its a bugger of a job I
know.)
The second thing that I found a problem was in duplicating the data streams in TradeStation. In your videos that you created when you show the
results on TradeStation you skip over the process of setting up the data streams. OK I know that is not really anything to do with GSB but for someone
like myself (and the others that are new to TradeStation) it would simplify the process. Especially frustrating was not being able use symbols because
they were either wrong or not subscribed to. I know the problem lies with TradeStation but it would be a relatively easy to incorporate the process in
a video.
A useful guide would be what you consider to be upper and lower limits for specific settings.
I found that the lack of the ability to save and retrieve previously created system results a bit annoying and cannot understand why this in not part
of the system. If is is then I could not find it.
I appreciate that GSB is still in its infancy and so I will only mention a couple things that I noticed that made it difficult for me to use.
Start, Pause, Stop buttons. They do not change colour or highlighting when entered into so one does not know if they are paused started or not.
Initial Capital appears to be missing.
Help file – None
OK that’s it.
Having said the above I will probably be purchasing the software as it does create very good systems so well done on that front.
|
|
|
admin
Super Administrator
Posts: 5069
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Thanks for your comments. I admit the core initial group behind GSB are very experienced traders, and the new comers to trading is not so well catered
for. GSB documentation also has assumed basic knowledge of TS. The good news is there are two separate people working on courses / docs. One of these
will be released to the GSB forum. I think both are aimed at people without the TS / MC prerequisites.
You can ask your questions here too.
http://www.trademaid.info/forum/viewthread.php?tid=4
My last video showed the setting up of data streams, but before too long we will make GSB code time zone agnostic. That's also going to help many
users.
https://www.youtube.com/watch?v=UNAEy1wYgho&t=19s
what settings specifically are you asking about?
"A useful guide would be what you consider to be upper and lower limits for specific settings. "
As a rule if you must touch settings, only change a little at a time, else you end up in a deep hole. A number of settings has limits imposed on them.
Systems can be saved and loaded by right clicking them.
There is a fixed capital mode if you trade shares / fx
Agreed on the stop buttons. We will improve that but there are much more pressing features needed.
Thanks for the positive comments. Bottom line GSB works well.
Help file will happen eventually.
|
|
|
admin
Super Administrator
Posts: 5069
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Later note. If you go to the top menu process, you can clearly see stop pause start with options greyed out depending on the current working state of
GSB
Halfax NZ is one of the people working on the course for GSB.
You might like that as its close to home.
I can email you what they have done, but its biased for Multicharts from memory.
|
|
|
admin
Super Administrator
Posts: 5069
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
This is how the buttons work, agreed would be nice if the red green yellow change color too.
|
|
|
boothy
Junior Member
Posts: 54
Registered: 21-5-2018
Member Is Offline
Mood: No Mood
|
|
Crazy Good results!
Hi, I have just started the GSB trial, for my first attempt I did a run on NG 30m and stopped the optimization after it produced the first 100
systems, then chose the best 3 equity curves and did walk forward analysis on them. One of them scored 100% for all of the WFA metrics!
Is this common?
Have I missed something?
It almost seems that the system has performed too well and too stable in the WFA.
Just curious if this happens often or did I just get lucky on my first run or have I missed something and its totally curve fit!
Cheers!
 
|
|
|
zdata
Junior Member
Posts: 9
Registered: 28-2-2018
Member Is Offline
Mood: No Mood
|
|
GSB is good for ES and NG system generation. 100% Walk forward result eliminates the possibility of curve fitting in my opinion. I will also run it
through tradestation walk forward optimizer just for the heck of it not required though. I yet to find 100% system in ES. NQ and CL are the hardest
one.
|
|
|
![[*]](images/davis/default_icon.gif){kind=icon}
