| Pages:
1
2
3
..
17 |
andrew.gibbs777
Junior Member
Posts: 17
Registered: 11-5-2017
Member Is Offline
Mood: No Mood
|
|
Wish List
Loving the software Peter. Here is my wish list
1 - Top Priority
2 - Must Have
3 - Would be nice.
1 - Top Priority
- MT5 Export Code (C++)
- Pre-Installed Data (from MT5)
- Save system results in format that can be directly imported into Portfolio
Analyst (without the need to export system to software first).
Functionality
Time of Day Filters
Price Pattern Filters (example: High>High of yesterday)
Range Based Filters (Decile Ranking System of Previous Bars/XBars Range relative to Average)
Stop Loss
Profit Target
2 - Must Have
Add additional Exit types:
To be selected manually at beginning before running GSB
- Exit after X Bars (Market)
- Exit on Open (Close) of Day After X Bars (Market)
- Exit on Open (Close) Of Day +/- Offset After X Bars (Stop)
- ATR/Range Trailing (Chandalier) (Stop)
- Bollinger Band Trailing (Stop)
- % Trailing (Stop)
- $ Trailing (Stop)
- Exit Long at Lowest Low (Highest High) in Trade +/- ATR (Limit)
- Oscillator Value >/- X (Market or Limit). Example: If RSI > X then Exit
Long Next Bar at Market or If ...(RSI>X) then exit long at Open (of day)+Offset Limit etc.
- Oscilator Target (eg Upper Bollinger Band) etc. (Limit)
- Testing on multiple primary data streams.
3 - Would be Nice
Intrabar Entry Types. (Stop and Limit)
- Next bar Open+/-Offset Stop etc...
Will continue to brainstorm.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
The ts export is going to go though a fair few changes, esp when adding filters / truism's. This will mean each time that happens, mt5 will also need
updates.
So I need to wait until the next stage is done.
Results into Portfolio Analysis should be do-able but some months away. Its a good idea.
Simple form of exits will be not to far away. Its good you say to do before running GSB. To do them Genetically is going to have to much risk of being
like Adaptrade builder. Redundant logic.
Multiple primary data streams is working, but not in a basket sense- Where one system can trade mutiple data steams identically. GSB can build systems
on say es.15,es.30,emd.30 as data1. But the system will only be trading one symbol when its finished, but systems on all the symbols will be made.
Intrabar stop, limit is planned but quite a few months away.
Thanks for the good feedback
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Exits of stop and profit target, + after x minutes have been done. Other exits will be done but there are more urgent things like Truisms and
improvement to WF
Got another user request for portfolio analyst export too.
|
|
|
andrew.gibbs777
Junior Member
Posts: 17
Registered: 11-5-2017
Member Is Offline
Mood: No Mood
|
|
In addition to Play, Pause and Stop can you also add a "Reset" button which clears everything so we can start again.
|
|
|
crazyhedgehog
Junior Member
Posts: 14
Registered: 9-5-2017
Member Is Offline
Mood: No Mood
|
|
+1 on export of either (1) Unique Strategies or (2) WF Strategies tab to Portfolio Analyst. This is for me the last missing piece of the puzzle in
terms of functionality.
Alternatively, if simpler, could we add a button/output tab to calculate correlation/negative correlation/positive correlation between strategies in
the Unique Strategies / WF Strategies tab? This would allow a workflow (1) Build strategies > (2) Run WF on strategies that look interesting > (3)
Calculate correlations to pick a set of strategies for a portfolio. I imagine it would be simpler to implement this directly in GSB rather than
creating and exporting files for Portfolio Analyst, and also much faster in terms of the workflow.
Thanks!
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
"+1 on export of either (1) Unique Strategies or (2) WF Strategies tab to Portfolio Analyst. This is for me the last missing piece of the puzzle in
terms of functionality."
Good ideas.
Both are on the to do list. Fix WF (possibly now done) and fix TS <> GSB at times are still the top focus. After that we should see these 2 features
soon.
|
|
|
Gregorian
Junior Member
Posts: 97
Registered: 23-5-2017
Member Is Offline
Mood: No Mood
|
|
+1 on Time of Day filter. Starting trading at midnight by default isn't good for thinly-traded-during-off-hours instruments such as CL.
|
|
|
gmoney
Junior Member
Posts: 14
Registered: 8-5-2017
Member Is Offline
Mood: No Mood
|
|
Be able to change the number of indicators used. Currently it is 3. I would like to be able to use 1 or 2 to reduce the number of degrees of freedom
(and number of variables to optimize).
|
|
|
Gregorian
Junior Member
Posts: 97
Registered: 23-5-2017
Member Is Offline
Mood: No Mood
|
|
One more thought: Make the Time of Day filter start and close settable parameters in the strategy that can be optimized in TS or MC. Right now the MOC
figure on the instrument settings is a hard-coded part of GSB and not optimizable.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Time of day start is going to happen, and maybe last trade of the day. ie don't take new entry after 14:30. Both as additional secondary filters.
These are good ideas that might be a month away.
As for adjustable MOC. Anything on indices/ gold 1500 is the best time. Hence I didnt feel the need to optimize this.
GSB is going to work best when you avoid things that seldom work. Its going to wast its genetic time on red herrings.
Even energies, bonds etc if your day trading, the end if USA day time is best.
Unless you can present a strong case this wont change.
Same applies for stops. Stops dont make or break a system and so you can manually optimize this in TS separate to the other parameters.
I dont want to end up having the issues of Adaptrade Builder. Too much rope and you can hang yourself.
|
|
|
Carl
Member
Posts: 342
Registered: 10-5-2017
Member Is Offline
Mood: No Mood
|
|
Hi Peter,
I would like to see the GSB settings I used as comments in the Tradestation script, like:
fitness function, stoploss, profit target, time exit ("minutes"), commission
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
We can do that down the track. I also want things like gsb build number too. Stops and PT should be in the code if it was used.
|
|
|
Carl
Member
Posts: 342
Registered: 10-5-2017
Member Is Offline
Mood: No Mood
|
|
That's true, but after optimizing the values in TS, you don't know what the build settings for stoploss and profit target were.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Don't think what you say is correct. Changing workspace variables doesn't change code. What you might be saying is you want the pre wf variables and
the post wf. That might be nice to include. Will see what we can do.
|
|
|
Carl
Member
Posts: 342
Registered: 10-5-2017
Member Is Offline
Mood: No Mood
|
|
I suspect when in GSB I set stoploss to 500 USD and profit target to 300, GSB comes up with different strategies than when I set stoploss to 3000 USD
and profit target to 2000 USD?
So I suspect the value set for stoploss and profit target influences which strategy the user chooses in the end?
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
It will make some difference if the stop and pt are so tight. Ive seen one user to that with good results, but it wasn't confirmed by having the code
back in TS with intra-bar order generation turned on.
GSB cant tell in some cases if a stop or PT is hit when they occur on the same bar.
|
|
|
Carl
Member
Posts: 342
Registered: 10-5-2017
Member Is Offline
Mood: No Mood
|
|
Suggestion:
Nice:
Maybe a good idea is to add the variable names in the tab walk-forward so it is easier to see what variable the value belongs to.
Cosmetic:
And maybe set the column width of the parameters automatically so you can immediately see how all the parameter values changed.
At the moment the column most of the times is too small to see all parameter values immediately.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Hi Carl, I think some of the column widths have been done on today's build. It was a lot of work to get systems to save and load, so less Gui changes
compared to other builds.
|
|
|
Carl
Member
Posts: 342
Registered: 10-5-2017
Member Is Offline
Mood: No Mood
|
|
Hi Peter,
According to your documents and notes your development process using GSB looks like this:
1. run GSB
2. select good looking strat in GSB and perform WF in GSB
3. export code to TS
4. WF opt in TS
5. WF analysis in EWFO
An very nice feature would be to be able in GSB to perform the (TS) optimization process and generating the (TS) WF files for EWFO.
After the validation in EWFO you can copy the strat code to TS.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Hi Carl,
That's in interesting idea. It could be done but Im thinking more of put increasing amounts of ewfo into GSB. Possibly even more WF features in GSB
than WFO.
Ie two pass WF, and auto fitness detection. It might even happen that both programs get this added.
Is there any advantage of making GSB make the TS files for EWFO. Im feeling pressure from people to reduce the need for EWFO and TS.
|
|
|
Carl
Member
Posts: 342
Registered: 10-5-2017
Member Is Offline
Mood: No Mood
|
|
Hi Peter,
It takes hours and hours in TS to optimize only one strat.
GSB is very fast in developing systems. So I suspect GSB is also much faster in optimizing than TS?
And it saves time and effort not having to copy strat code from one app (GSB) to another (TS).
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Thats true but its best to manually optimize each input and look for any indicators that are not valid.
Most system builders do not have this in there methodology (if they even have a methodology) and its going to help OOS success I think.
I think GSBsys1 and GSBsys2 both had this issue in their original code. Its hard, but possible for an automated process to pick this.
Using the GBB (FFC) functions can make dramatic differences to the speed of optimization too.
Thats also why its nice to have a decent i7 cpu, possibly over clocked.
Over time GSB is going to automate in increasing measures the entire system building approach.
|
|
|
Carl
Member
Posts: 342
Registered: 10-5-2017
Member Is Offline
Mood: No Mood
|
|
You could add a feature in GSB to optimize per input?
1. GSB - manually pick a strat
2. choose optimize
3. GSB gives a dropdown menu with the strat inputs
4. user pickes input1 + optimize
5. analyze the results
6. go back to step2 and pick another input to optimize
This would still save a lot of time and effort.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
I like your proposal.
Everything you said could be fully automated and is a good idea. It might be hard to automate finding a redundant indicator unless its very strongly
not effective.
I think that maybe should be done before the second pass WF.
Im going to put this on the Todo list, but it might be a month or 2 away.
Bug fixes then truisms are the short term focus.
|
|
|
Carl
Member
Posts: 342
Registered: 10-5-2017
Member Is Offline
Mood: No Mood
|
|
And if the user is optimizing a weight input and a value of 0 gives the best performance, show a message "possibly redundant indicator"
|
|
|
| Pages:
1
2
3
..
17 |
![[*]](images/davis/default_icon.gif){kind=icon}
