| Pages:
1
..
9
10
11
12
13
..
47 |
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
I hope to have a very significant improvement to out of sample results in the next build. 52.50
More tomorrow. Testing it now.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Results are looking good, but not finished all my walk forwards.
signpower I like best with multiply, but signlog is also good.
Both are better than the previous modes
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Here are some results,
also shown is the results from https://trademaid.info/gsbhelp/Provingthemethodology.html
Improvements due to signpower function, and tweaks in methadolgy
User Bruce has also verified with his own testing.
|
|
|
Bruce
Member
Posts: 115
Registered: 22-7-2018
Location: Auckland - New Zealand
Member Is Offline
Mood: No Mood
|
|
Yes, this is certainly delivering some encouraging results and consistently so. I'm now running these newer 'tweaks' across other indexes, bonds, EC,
etc to see what those builds deliver.
|
|
|
John62
Junior Member
Posts: 23
Registered: 29-9-2017
Member Is Offline
Mood: No Mood
|
|
If I do not want to prove the method but use it (to keep it simple) can I perform the following steps:
1. use all dates, Nth 1, NoTrd, Period 80. So 50% of data is OOS.
2. Select the best indicators, using 2 indicators.
3. Apply to optimization setting.
4. Buils 50k systems with the setting as described.
5. Change Nth to all.
6. Sort on NP / DD and put the top 2000 in favorites A.
7. Sort on Pearson's and remove the bottom 1000 from favorites A.
8. WF these favorites A.
9. Sort on Anchored stability> = 40.
10 Analysis and pick the systems I like.
Or am I missing something this way?
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Hi John
what you say is good, but missing some details, and Im updating this all later in the week.
3.1 change to 3 indicators
6.0 use the macro. Sort on fitness, pearsons 0.98 pf 1.5, TOP 250
7.0 Delete this
I have spent many months and have about 9 computers / servers that spend most there cpu working on this.
Hence there are some small tweaks that give very reasonable improvements.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
The docs are updated to the current version, though not as through as I would like.
There is now more programing hours per week going into GSB, so that makes it harder for me to document.
Here is how I got a 50% increase in out of sample results with the identical data set from last month.
https://trademaid.info/gsbhelp/Updates.html
Whats in the pipeline is Artificial intelligence in walk forward.
Typically there is a lot of merit to optimizing each input manually in TS. Its not essential but sometimes things come up.
For example if indicator1*weight1 gave best results when weight1=0, you know you've got a redundant indicator1.
Well AI should be able to tell you this.
So this is going to roll out slower over time. There will be a new WF called AI.
It will start by just doing some graphs, then later builds of GSB will draw its own conclusions on the graphs.
|
|
|
Bruce
Member
Posts: 115
Registered: 22-7-2018
Location: Auckland - New Zealand
Member Is Offline
Mood: No Mood
|
|
Peter, These latest updates are working great!
I just wanted to share a really simple build that is delivering solid results. Here I've just used the ES30 with the NQ30 and YM30 (instead of
ES29,ES31), followed the same process you've outlined in the help using your Macros and the results can be seen. I know you are likely doing better,
however, I'm just a hack and let's not forget you're the pro! 
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Hi Bruce,
there is a massive amount of things to experiment and try.
VS and VSS and random noise, so its good to see other things being tried.
Those results however are well down from the current method.
Spending all day on GSB is also highlighting how much we dont know, and assumed.
What I like is our assumptions can now be tested. Plus being with a great community who all have there own little tweaks.
One person can do little. A group of people can do so much more.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Here is the first system I chose using some of the new methodology. It also has a slightly different entry mode that I feel is slightly better.
There is one more entry mode that I have still to try
results out of sample since 28/3/2019
Includes $4.8 and $25 slippage and commission round turn
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
In life, its good to ask the right questions.
My limited knowledge of a field where I have no expertise is, 50% of knowledgeable people think dna is like a language. The other 50% think it is a
language. There is a lot of implications to this.
Again with walk forward, People feel strongly over anchored and rolling WF.
People with much more expertise than me feel strongly about opposite opinions, and the answer I dont think is purly a yes no answer.
As for do we optimize and or build separately long and short...
I have been experimenting a litttle with building long systems on ES, and building a short system. Then put both systems on the same chart. My limited
research using market validation techniques showed this was successful OOS compared to symmetrical
system.
What I tried today was to optimize the same system long, and then short. Adding both on the same chart.
It was very successful. It also solved the annoying issue that many systems want a different parameter set to cope with 2018/2019.
I had a earlier system that failed some of my checks late in the testing process. It did not help this failed system.
So the degree to how good this is, remains to be seen.
The next release of EWFO will have the ability to do a long only, and a short only WF.
This is not at all the same as a long & short at the same time wf, where you display the long metrics and display the short metrics seperately.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
docs for the new ewfo feature are here
https://trademaid.info/gsbhelp/LongShortWalkforward.html
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
Hi Peter,
Many thanks for info reg ES Wf long and short.
1: I have seen your comments in forum about long and short separate on same chart, my first thoughts was.. is this setup handled the same in TS and
MC, according to MC support, two separate signals on same chart is seen as one and same signal. I am thinking what difference would it be two have two
charts and one short on one chart and one long on another chart, compared to having both on same chart ? how is this handled in TS, mc? Your
observations?
2: In regards to your breaktrough, shall i understand it that you built one signal and then separated long and short, and WF each one separatetly ? or
did you built one long and one short, then WF them separately in TS?
3: You also mentioned that year 2018 and 2019 uses different Prm than previous years, i have always built systems with data to 2015 BUT then used data
to 2018 02 28 when doing my WF, this makes me think that i should try to also include the most recent data ie to today date, is this your way to WF,
to also include data to todays date in general ?
4: My take away from all this is that, using separate BUILT systems for long and short could be most likely optimal and then also use use all data
possible in the file when WF, and that you observations show that long and short most likely benefits from WF separately. Correct?
5: On a sidenote, i was going to make a test to build systems for BTC cme future contract,
https://www.cmegroup.com/trading/equity-index/us-index/bitco...
However i cant figure out what settings to use in Contract list
Ticks: ?
Pnt Val:
Digits:
Can you please help?
6:
When building systems long and short separate for ES, i find that its much harder ie higher c/us.
Do you use same settings for long and short separate, as when building combined ?
Same optimal indicators and same Opt settings, SF etc?
Many thanks again Peter for your kind support
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Hi Daniel,
More from me later
1) This will give long from one code and short from another.
Likely the tightest stop will apply to both codes
2) The next GSB build can do a wf thats long or short only
EWFO next build can do this, and combine reports
I am not certain of how well this concept works, esp on 2018 on ES which is a volatile but hard year
6) I used pearsons 0.91 on training, not the usual 0.95
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Hi Daniel,
regarding 3) Im not decided on this. Its an unusual situation. But after 2008, it was really bad practice to use the last year as the extreme range
curve fitted the system to only do well in extreme range. This was not obvious to me at the time, but is now.
4) I think so, but im still studying this. This difference on 250 systems long and 250 optimized short is a mild improvement.
However my philosophy (from jamesclear.com) is many consecutive small improvements lead to big cumulative improvements.
5) Will look at early next week.
|
|
|
saycem
Junior Member
Posts: 48
Registered: 13-7-2018
Member Is Offline
Mood: No Mood
|
|
I have been building on ES 29 30 31 to 6/2015 using the new methodology settings.
Running the BestIndicator Macro and then building 30k systems with top indicators, but I have not been able to find a system anywhere near the perf
metrics of what Peter posted above. The Max Net Profit I could get was about $77k for example - Peters system >$132.
My stats for the first run had low deg -2.26 but after running again with Opt Indicators -22.0.
I'm wondering if anyone else has had same challenges? I've re-run a couple of time now. Posting my settings if anyone can see something incorrect?
    
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Hi Saycem,
A critical thing is your contract moc is 14:30, not 1500
Your indicator set differs, not sure if its 8 or 38 from what you posted
Under general. I am using 0,0 not the 1,1 you have.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
New ewfo is out with option to WF long, and short and then combine.
The combine is not quite as complete in reporting but still very useful.
Using EWFO in this mode will likely give different parameter set for long and short.
see https://trademaid.info/forum/viewthread.php?tid=123 for the exe
See this url for the documentation on it
https://trademaid.info/gsbhelp/LongShortWalkforward.html
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
Thanks Peter, A thought... Lets say i dev a nice system from GSB for ES and after WF normal long and short together, you end up with stability scores
of 90ish... all great...
I then would like to test WF long and short individually... however now i end up with stab scores of 0 and 5 for long and then short.... so WF
combined stab scores great, but when separate i get higher NP with same DD, but stab scores are really bad (something that you would have rejected
normally).
i then develop long ans short systems individually and end up with long and short systems that has stab scores or 90-100 for long and short systems
individually.. is this not better?
How to think here? about stab scores in WF when doing WF for a system that is both long and short but then WF separate for long and short.
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Hi Daniel,
there are lots of issue here, and much more control when we get GSB 2.0
Check your fitness is on np*at
Was the files in ewfo from gsb (ewfo wf) or ts?
GSB doesnt always give the best parameter range. ie SF closeD way too big.
I would be happy on es to make it 13 to 16. In gsb 2.0 the plan is to have much more control over these things.
|
|
|
Daniel UK1
Member
Posts: 470
Registered: 4-6-2019
Member Is Offline
|
|
Quote: Originally posted by admin  | Hi Daniel,
there are lots of issue here, and much more control when we get GSB 2.0
Check your fitness is on np*at
Was the files in ewfo from gsb (ewfo wf) or ts?
GSB doesnt always give the best parameter range. ie SF closeD way too big.
I would be happy on es to make it 13 to 16. In gsb 2.0 the plan is to have much more control over these things. |
Hi Peter,
It was not so much an issue, more how to think and perhaps your thoughts on the dilema, perhaps i was unclear, sorry for this.
What i mean is that when system developed for both short and long in same strat, WF you have good Stability metrics like 100... BUT when WF long and
short separat, stability scores are very low like 5...
In this instance, would it not be better to developed short and long separate, and just pick systems that have optimal Stability scores for long and
short separate?
I have not done any WF in EWFO, just GSB...
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
Hi Daniel,
I would look more closely at why the stability is low, and also look at stability-c (coarse)
The logic with coarse is build in some tolerance.
ie a length from 99 to 100 is 1% and irrelevant, but will degrade the normal stability just as much as going from 99 to 50
Coarse stability has under app settings a tolerance value
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
There is significant but not complete updates to the docs, which shows the power of GSB, and various architecture types.
Its aimed more at experienced users rather than new comers.
You may need to refresh your browser to get the updates.
A number of new features in the yesterdays release is also explained.
https://trademaid.info/gsbhelp/SFIndicator.html
https://trademaid.info/gsbhelp/WFparamaters.html
https://trademaid.info/gsbhelp/Weightmode.html
https://trademaid.info/gsbhelp/SignMode.html
|
|
|
admin
Super Administrator
Posts: 5060
Registered: 7-4-2017
Member Is Offline
Mood: No Mood
|
|
At long last, I publish my default settings I am using for testing.
All files including data files are included, but not all the gsb exe,dll, xml files.
Included is gsb.config.xml to get GSB into admin mode.
There may be some settings that need a tweak, so consider this the first revision.
There are 4 ES files with 5 ticks of noise, 4 ES files with 10 ticks of noise
I think you can put this into the folder of your choice. (All paths relative, not fixed)
200mb in size zipped up.
My top 11 indicators are selected, and these are better than any other method Ive found. However sorry to say, im not sure how I found this group.
All macros used are included.
In summary you build 50,000 systems. Dates are 2000 to 2015.6.30 nth 1 / 80
The manager then populates but with nth set to all (workers built them with nth no trade)
Then macro 1 selects the top 250 systems and walk forwards them.
Manually put the top 250 into favouritesD.
Then run macro3 to get the stats.
record these stats.
Then run the macro5 astab-c macro that deselects all favouritesD, and puts the top 90 astab-c
Then run macro3 to get the stats.
The record these stats.
go to favourites A
Then run macro2 which changes dates back to 2015.6.30
after this is complete, run macro6 which selects the top 91 vss systems of the top 250
Then run macro3 to get the stats
Shown below is a summary of my stats.
zip file here of all my files in private forum
https://trademaid.info/forum/viewthread.php?tid=241#pid4133
|
|
|
appengineer
Junior Member
Posts: 61
Registered: 8-4-2019
Member Is Offline
Mood: No Mood
|
|
Hi Peter - Macros 4 is missing, can you share it, Thanks.
|
|
|
| Pages:
1
..
9
10
11
12
13
..
47 |
![[*]](images/davis/default_icon.gif){kind=icon}
