Update to GSB methodology. A must read, the backpacker and the Art of war by Sun Tzu

Quote: Originally posted by george_holland

Hello, I'm looking at the methodology part now, but I'm a bit confused; I find different video's and docs about the methodology. What is the most recent methodology we should follow?

Quote: Originally posted by admin

@portfolioquanttrader2020 These should be in your GSB FOLDER. macro c:\gsb\Data\Settings\Macros\NQ.YM.ES\nct3-300_build_p0.9_AT+stats_is2017.2_2018.2_2019.2_2020.2_2021.2-NoStartDates.gsbmacro Optsettings C:\GSB\Data\Settings\Optimization\Manager\nq-800-1500_30minBars_75i.gsboptset

Update for the last 1 or 2 weeks
Ive been working on GSBsys1Es but on the NQ
very unusual for me to spend so much time one one system.
I found something not easy to notice while manually optimising the system that is both very good and unusual.
While im would be happy to trade and release the code, i feel I have not yet reached full potential on the idea.
Here is the nq system with and without one little tweak I made
Incidently, thats why i was looking for I9 servers to benchmark. The painful speed of TS can be much improved with a better CPU.
unrelated, but whats best CPU for GSB is likely not the best for TS.
One of the numerous factors is TS can use a max of 32 cpu cores, & possibly only 1 cpu core on a dual cpu macine.
Shown here is with and without this new filter.
GSBSYS1nqv2 will be a paid release.

Quote: Originally posted by Zwijnhond

Which settings would you recommend to tweak / test for other markets?

Hello
I need to know where the way to interpret this table is explained
Cheers

I now have some objectivity if we should trade fed interest dates etc.
This is my beta concept, and Im open to alternative view points.
I am testing using about 20 or so systems. Quite a number are related systems. IE GSB1ES series.
So I dont want to curve fit on one day trading system, but on the basket of them.
There is > 10,000 trades and I only will consider a news event that has 200 trades. Too small a sample is not useful statistically.
I will stop trading on events that have got really poor NP, np/dd or PF
Note some of these are profitable long term, but higher risk and lower reward.
Now you can rightly argue that what I have done here is said with no out of sample period, but I feel its likely to be valid

It is possible by date filters as well, to do a in sample test as well on PA.
But I argue the concept of not trading fed interest dates is not a new one. Numerous uses have expressed this to me. Traditionally I would trade all days
except 1/2 trading days. Bottom line is no one will get this right all the time, but the volatility and choppiness on these days is likely to be high for some time.
I will likely add this code into GSB systems as a default, with inputs to exclude any 1/2 days
ie dontTradeDate20221125);

another GSB community project
We all have our minds holding to a set of beliefs that we are sure is true. Applies to trading and life in general.
I think it was Mark twain who said, what you dont know wont kill you. Its what your sure is true and aint that will.
Well my firm beliefs are wrong again. I did a lot of testing and found 30 min bars to be the best.
What I may not have done is tweaked the start time so all bars are of equal time value
ie 830 to 1500 is 13 30 min bars
if we go to 29 min bars we are short 13 min, so the session time should be 843 to 1500 etc
Well I have done the research on ES and found 30 min bar was the 26th best interval, and 15 min was the 58 th best bar interval.
Have completed nq too, but made a mistake so have to redo it.
So send me your share key, and I will add you into the cloud dedicated to this project, and will send you the full results of nq and es testing
If you have not given me your worker share key in the past, make it youremailadress.1234 were 1234 is a unique random number of your choosing

Quote: Originally posted by admin

Bar interval tests have been done on es ym nq rty with secondary fitler highlow3 Closed not normalzied secondary filter had been done on ym. more too come. thanks all results in pcloud

Quote: Originally posted by admin

@cico, I may have a different list and its not critical. Likely best to just use the full set for gsb purchasers. Beta indicators may technically work, but not give anything unique or as good as default indicators & a few dont match GSB to TS. Over time I move indicators that truely give value into GSB purchasers list. Whats in excel is always less that what you started with. If you then start with only whats in excel.. go through the process again you have a smaller list. I repeated this process numerous times to get a small set each time and surprisingly it was a total waste of time.

Peter,

https://trademaid.info/gsbhelp/Maximumparameterduplicates.ht...

1- Could you please explain "Maximum parameter duplicates" more?

in the recent community project we're using 2, but in Builds we're using 0?

-----

2- Could you please explain SF-Indic. Params & SF-entry level and how they relate regrading entries Long & short? see attachment

3- TF Parameters? so in attachment 48 ticks above prev day close?

Thanks

Energy

Quote: Originally posted by admin

@us works poorly. Energies likely ok, but the secondary filter to use is one of the first things to look at. Likely closedlessclose normalized is first to consider. It's been years since I looked at the other markets so don't remember much more

Quote: Originally posted by portfolioquanttrader2020

I made these videos in Spanish, but you can translate subtitles: https://www.youtube.com/watch?v=RJdCDpZeB0I&t=3s https://www.youtube.com/watch?v=vhI30fRH5Wo&t=590s https://www.youtube.com/watch?v=Wdye9_1rPkk&t=375s If you need help, I could help you

Quote: Originally posted by REMO755

Hello, Select whatever systems gave the highest stats from the top 250, or top astab-c 90, or top vss 91 Where do I find the Macro to do this? top astab-c 90, o top vss 91, ¿qué significa esto? Thanks.

A good tip that will serve you well, is... take all systems after your process (that needs to be sound) from A, create families from all... WF all systems in family 1-5.. pick from families that have most systems with WF stab over 70 preferably ... you want the families with highest stabilty... all systems should have similar metrics.. then review WF runs for all sytems, and look for nice and stable WF per year OOS... then test systems, in similar markets, you want profitability, not much but positive, then take same system run on random ticks generated data, 5,10, 15, see how they cope... make sure you have saved at least 6-12 months of unseen data.. so when you transfer to platform, you have 6-12 month of unseen data, that hopefully your system passes... so forget the existing macros, understand what the macros does though and the gsb process from Peter, then create you own macros, dont use other peoples macro, no bueno

Quote: Originally posted by Daniel UK1

Quote: Originally posted by REMO755   Hello, Select whatever systems gave the highest stats from the top 250, or top astab-c 90, or top vss 91 Where do I find the Macro to do this? top astab-c 90, o top vss 91, ¿qué significa esto? Thanks. A good tip that will serve you well, is... take all systems after your process (that needs to be sound) from A, create families from all... WF all systems in family 1-5.. pick from families that have most systems with WF stab over 70 preferably ... you want the families with highest stabilty... all systems should have similar metrics.. then review WF runs for all sytems, and look for nice and stable WF per year OOS... then test systems, in similar markets, you want profitability, not much but positive, then take same system run on random ticks generated data, 5,10, 15, see how they cope... make sure you have saved at least 6-12 months of unseen data.. so when you transfer to platform, you have 6-12 month of unseen data, that hopefully your system passes... so forget the existing macros, understand what the macros does though and the gsb process from Peter, then create you own macros, dont use other peoples macro, no bueno

@portfolioquanttrader2020
can you remind me how this works?
We had it on a gold system years back and I remember it was very problmiatic.
I think it gave an error until chart was turned on a second time

you can see

Pre-selected indicators

Quote: Originally posted by admin

New methodology video Info and video at the start of this thread is obsolete now the two pass method here using green/green applies to all markets. https://trademaid.info/gsbhelp/Crudeoilfastrevisedmethodolog... Simulated data, 29,30,31 minute bars. Simulated data, 26,30,34 minute bars. Simulated data, 29,30,31 minute bars. ES, IDX, SPX Until more recently I was not aware that GSB could work on more markets than I originally found. There were 3 keys to this. The first one was to get the correct session times. The second was the finer details of the settings, and the 3rd was pure CPU grunt. (Except natural gas) One of my favorite jokes is a joke about knowledge extrapolation. Smart people assume that knowledge in one field, amounts to knowledge in another. GSB users including myself have also done some dangerous knowledge extrapolation. A doctor, a lawyer, a little boy and a priest were out for a Sunday afternoon flight on a small private plane. Suddenly, the plane developed engine trouble. In spite of the best efforts of the pilot, the plane started to go down. Finally, the pilot grabbed a parachute, yelled to the passengers that they had better jump, and then he bailed out. Unfortunately, there were only three parachutes remaining. The doctor grabbed one and said "I'm a doctor, I save lives, so I must live," and jumped out. The lawyer then said, "I'm a lawyer and lawyers are the smartest people in the world. I deserve to live." He also grabbed a parachute and jumped. The priest looked at the little boy and said, "My son, I've lived a long and full life. You are young and have your whole life ahead of you. Take the last parachute and live in peace." The little boy handed the parachute back to the priest and said, "Not to worry, Father. The 'smartest man in the world' just took off with my back pack source: http://www.jokes4us.com/morekiddiejokes/alawyeronaplanejoke....

This deserves a round of applause, it's a very good joke! ja,ja,ja

I have learnt a lot this week.
Spent over a week applying simple filters to existing systems with good results
What I am thinking...
some markets revert to the longer term mean. This likely means gsb systems do not change, but have a filter that looks at slightly longer term trend and expect the short term price to go closer to the long term price.
What ive learnt should be able to be applied to new indicators and filters, but it still requires a bit of thinking & work on my side
Enclosed as an example of a filter. Note zero slippage was used, so this wrongly bias's to higher trade count

Quote: Originally posted by Daniel UK1

Quote: Originally posted by REMO755   Hello, Select whatever systems gave the highest stats from the top 250, or top astab-c 90, or top vss 91 Where do I find the Macro to do this? top astab-c 90, o top vss 91, ¿qué significa esto? Thanks. A good tip that will serve you well, is... take all systems after your process (that needs to be sound) from A, create families from all... WF all systems in family 1-5.. pick from families that have most systems with WF stab over 70 preferably ... you want the families with highest stabilty... all systems should have similar metrics.. then review WF runs for all sytems, and look for nice and stable WF per year OOS... then test systems, in similar markets, you want profitability, not much but positive, then take same system run on random ticks generated data, 5,10, 15, see how they cope... make sure you have saved at least 6-12 months of unseen data.. so when you transfer to platform, you have 6-12 month of unseen data, that hopefully your system passes... so forget the existing macros, understand what the macros does though and the gsb process from Peter, then create you own macros, dont use other peoples macro, no bueno

@daniel
here is gsbsys2nq
Only a filter is being optimized, no other settings are touched, so perhaps not a perfect example
Basically there is a 18% reduction in the amount of trades and a $500 in total increase in NP
equity curves are very parallel
Note that in run 2 and 4 only, the input value of 0.2 made this filter pass all trades
but when the filter kicked in in most other runs 0.06 was a fairly consistent value
atr value of 0.42 and above pass 100% of all trades

The filters section is going to grow a lot over time.
Here is a really poor nq system with one of the new filters added
This gives me great confidence in how good the filter is.
Crap system and all metrics get better both long and short.
This is in 65.75 build.
Im not in a hurry to release the build as the entire cloud and all users will need to upgrade due to version incompatbiltiy.
Im waiting for more updates before I release it

Quote: Originally posted by admin

@daniel here is gsbsys2nq Only a filter is being optimized, no other settings are touched, so perhaps not a perfect example Basically there is a 18% reduction in the amount of trades and a $500 in total increase in NP equity curves are very parallel Note that in run 2 and 4 only, the input value of 0.2 made this filter pass all trades but when the filter kicked in in most other runs 0.06 was a fairly consistent value atr value of 0.42 and above pass 100% of all trades

@TwntySQ, these are the dates im using.
I think yours are not correct, but im not certain what values are in the trial build

You're correct here Peter, I must have configured the dates period. My bad

the short cut method will be removed.
This is the weakness
when (not if) we get a bad set if indicators in a test, it has a really significant effect on all results.
If we use the normal method, with 100 tests and filtering results to 2 standard deviations, a bad indicator set has no effect
Here are the two lemons in 28 indicator tests, 25 tests on each indicator.
This method will get removed from the documentation shortly
https://trademaid.info/gsbhelp/ShortcutMethodexplained.html