yes, it automatically writes them into the code.
does that mean that v11 will always use the most optimal variables?
no, it just means that after you’ve done your optimization and found the values you want to use, you don’t have to insert them into the code, you can leave them in the optimization box.
NAS 2m HullSar V5.3L
Hello,
I use the lastest version of NAS 2m HullSar V5.3L, since 1/1/2021.
I test a new version with other paramatrer of 50K bars
With this new version the result are 19% better that the exceptional V3.5L
Here are the Itf and the board.
I make the test only on backtest no Walk forward
I test first on TF12; take the middle best (for exemple if the best hull 12 TF is in descending 181,182,183, i choose 182). On so on of the other TF and parameters
My question.
Every how long you make an reoptimization of your parameter ?
I think is usefull to create a new topics about discussion about reoptimization of this powerfull strategy.
Thank again @Nonetheless
Thank you for you comment
Hey, thanks for that – could be interesting.
As for optimization, there seems to be 2 schools of thought (although probably everyone does it slightly differently). Some optimize over 100k or 200k bars and repeat every few months to try to stay with whatever the market is currently doing.
I prefer the long term approach – optimize once on the maximum data and try to arrive at a set of values that has worked well enough under the widest range of conditions. If we accept that all optimizations are to some extent curve-fit, then i prefer to have the longest curve possible.
Hi, Where can I find V11 you’re talking about ?
When you load PRT you should be given the choice to launch v11.1 or v10.3
Seems like NAS 2m HullSar V5.3L is having a tough time during last weeks 🙁
When you load PRT you should be given the choice to launch v11.1 or v10.3
PRT of course ! I was searching a V11. itf …:)
I did a test 1M on DJ 1 min HULL-SAR 7.5. This was good during crisis but very average before and after (since november 2020). We also can observe the code is quite good in Long but not in short
I’m going to do tests on parameters to optimize them when outside crisis. I will share you results.
Here are the results in Long only,
I did test in 100k so from 07/12/20 to 17/03/21. The optimisation look very good in this laps time.
It’s still not so bad in 200k from 31/08/2020 and also from june 2020. But in 1M from 06/2018 to 03/2021, it doesn’t look good.
I think it’s a sort of code who need to be upgrade every 3/6 months.
Or maybe the 1 minutes is a too short time frame to be goods in any sort of market.
Wait your suggestions
Which parameters do you optimize?
Hello @winnie37,
I tried to optimize period parameters, SAR and RSI/Stoch parameters. I bigan by 15 min time frame, then 5 minutes and 1 minutes to finish.
I didn’t change stop loss and take profit.
You can find the itf files.
Curiously, i can’t run it in pro-order. It gives me a message who say me there is trouble line 18. I didn’t work on it and it same as before so i don’t understand why ot doesn’t work.
You also can find the original code in the previous attachments.
Enjoy
there is trouble line 18.
Phew! that took some finding! I persevered so I can give something back.
It’s due to use of variable name std at Line 65 (not line 18).
std is a reserved term for standard deviation.
I changed std to stl and deleted std in the optimiser and added stl in the optimiser.
Code then gets accepted by ProOrder.
could you post the ready to run version for testing in demo? thanks 🙂
LaurentBZH35 what timezone is your DJ-1m-HULL-SAR-v7.5-Lg-1.itf please?
I guess we all need to assume that timezone used in Algo’s are those of the Author or Editor … so in LaurentBZH35 case this will be UTC +1.
