I mostly use the Adaptive Net Indicators but also predictions. I keep all CSSA parameters constant for all timeframes so my training rarely takes more than a few minutes. I am very careful at not letting the optimization go all the way through because of overfitting.
Noxa indicators for Neuroshell
- Thread starter val2004
- Start date
- Watchers 71
I mostly use the Adaptive Net Indicators but also predictions. I keep all CSSA parameters constant for all timeframes so my training rarely takes more than a few minutes. I am very careful at not letting the optimization go all the way through because of overfitting.
My first post here.
Neuroshell seems not to be easy to work with. I got NS bridged with MT4 for datafeed and have NOXA CSSA installed.
Still playing around with it but in the end we all want a working system to trade live.
Hope we can get somewhere, it is a expensive piece of software xD ( atleast for me ).
Neuroshell seems not to be easy to work with. I got NS bridged with MT4 for datafeed and have NOXA CSSA installed.
Still playing around with it but in the end we all want a working system to trade live.
Hope we can get somewhere, it is a expensive piece of software xD ( atleast for me ).
windsurfing_stew
Member
- Messages
- 77
- Likes
- 1
Nick,
Be sure to ask any Neuroshell or CSSA questions you have in the private NST forum. Their support is really helpful.
Best of luck with it,
Stewart
Be sure to ask any Neuroshell or CSSA questions you have in the private NST forum. Their support is really helpful.
Best of luck with it,
Stewart
From what I can read in the NST forum, a majority of the Neuroshell users are ForEx and commodity traders.
windsurfing_stew
Member
- Messages
- 77
- Likes
- 1
Ummm, I think there is a pretty fair mix actually. From the NST Forum posts it looks like FX is about 1/3 to 1/2 of usage as a very wild guess. I only look at FX and Commodities myself.
arry,
I read couple of your pdf's on internet about neuroshell settings thanks you very much. I have a query if you dont mind. I have been testing a trading strategy but it seems I cannot select a stoploss in the exit rules. This has led to case where my trading system is having massive drawdowns. How can I select a stop loss in such a way that when I optimise a strategy its uses my stop losses used to define trading parameters.👍
I read couple of your pdf's on internet about neuroshell settings thanks you very much. I have a query if you dont mind. I have been testing a trading strategy but it seems I cannot select a stoploss in the exit rules. This has led to case where my trading system is having massive drawdowns. How can I select a stop loss in such a way that when I optimise a strategy its uses my stop losses used to define trading parameters.👍
Hi Amit,
I think you may use indicator into your protective loss into your trading strategy (based on entry price, floor price or trailing price). In addition, you should activate trailing stop interface on Trading Wizard, see Options to activate them.
Arry
www.ai-traders.org
I think you may use indicator into your protective loss into your trading strategy (based on entry price, floor price or trailing price). In addition, you should activate trailing stop interface on Trading Wizard, see Options to activate them.
Arry
www.ai-traders.org
I find it better to optimize my stops separately. This way they don't mess up with the selection of cycles.
windsurfing_stew
Member
- Messages
- 77
- Likes
- 1
Hi Mr Future,
A really nice looking trading system. What appears particularly impressive about it is that you have trained it on a rising market and it still performs on a falling market. This is unusual but commendable.
Just a few quick questions -
1) What do you mean 'two dimensional assessment'? Is that two inputs to the indicator, or what two dimensions of a probability density? Or two assessments of probability density? Is this probability distribution empirically derived, or you've used a Normal Gaussian distribution?
2) When you discuss the min-max? I take it you are recording a single sample point for each low or high on a CSSA cycle?
3) What do you mean by 'standard price range'?
Cheers,
Windsurfing Stew
A really nice looking trading system. What appears particularly impressive about it is that you have trained it on a rising market and it still performs on a falling market. This is unusual but commendable.
Just a few quick questions -
1) What do you mean 'two dimensional assessment'? Is that two inputs to the indicator, or what two dimensions of a probability density? Or two assessments of probability density? Is this probability distribution empirically derived, or you've used a Normal Gaussian distribution?
2) When you discuss the min-max? I take it you are recording a single sample point for each low or high on a CSSA cycle?
3) What do you mean by 'standard price range'?
Cheers,
Windsurfing Stew
Hi Amit_rwd,
As your private message query, please find here attached constant template (copy and paste in template folder). Insert it as indicator (in Custom Indicator) with two parameters Close and constant. Leave parameter 1 equal to Close as default and parameter 2 you can adjust as the constant value you want.
Limitation: the constant value is real number only between - 100,000,000 to +100,000,000 ( I think it should be enough). You can put real number with decimal point as you want (sample 0.000023) or integer.
Have a nice try, it is free (without any support and warranty 😆) but you can get commercial one as part of Advance Indicator Set 1 add on.
regards,
Arryex
As your private message query, please find here attached constant template (copy and paste in template folder). Insert it as indicator (in Custom Indicator) with two parameters Close and constant. Leave parameter 1 equal to Close as default and parameter 2 you can adjust as the constant value you want.
Limitation: the constant value is real number only between - 100,000,000 to +100,000,000 ( I think it should be enough). You can put real number with decimal point as you want (sample 0.000023) or integer.
Have a nice try, it is free (without any support and warranty 😆) but you can get commercial one as part of Advance Indicator Set 1 add on.
regards,
Arryex
Arryex,
I learnt that normalising data and feeding to neural nets increase the performance . 👍 However do you have any idea how one could normalise pivot points so nets can learn something from the pivot points..?
I tried to feed noxa into the neural nets but amount of trades are less and not convincing for me 1hr charts ( 200 trades in year) .. I used a parameter search to let find the best parameter.
😢
I learnt that normalising data and feeding to neural nets increase the performance . 👍 However do you have any idea how one could normalise pivot points so nets can learn something from the pivot points..?
I tried to feed noxa into the neural nets but amount of trades are less and not convincing for me 1hr charts ( 200 trades in year) .. I used a parameter search to let find the best parameter.
😢
Amit,
I think Neural Network will be depend on the following factors:
- Inputs that you feed (normalizing data/indicators are preferable rather than raw price data)
- Output that you intend to find (predict close/open, or percent change; classify the trade buy/sell/hold)
- Neural Network architecture, optimization methods, etc
- Data range used for training, test/cross validation and production/out of sample (percentage data range influence the net)
Previously I've made some mistake to select properly above factors and made my prediction /classification or feeding unnecessarily my net with wrong data.
Concerning the utilizing pivot point, why not? you may feed your net with pivot point values as inputs and future percent change of close/open as output. I would like to try them also.
As far as I know Noxa has limitation, i.e. it should be adjusted manually in order to matched with the market cycle (peak and valley of price data), as long as the frequency is matched that you will get profit from it. I received a statement from someone that it is difficult to use GA automatically to find the best noxa parameter. If you feed them to the net with different frequency from the market, then you feed your net with irrelevant data, you can not expect it will produce profitable trade.
Personally, I made some investigation to find what are the best output that can be used for my net, which inputs shall be selected, how, etc. I believe Neuroshell has capabilities to verify them. I prefer to return to basic/core idea how we can utilize NN, instead of trying all indicators without knowing what we dealt with (one of my previous mistake).
Arryex
I think Neural Network will be depend on the following factors:
- Inputs that you feed (normalizing data/indicators are preferable rather than raw price data)
- Output that you intend to find (predict close/open, or percent change; classify the trade buy/sell/hold)
- Neural Network architecture, optimization methods, etc
- Data range used for training, test/cross validation and production/out of sample (percentage data range influence the net)
Previously I've made some mistake to select properly above factors and made my prediction /classification or feeding unnecessarily my net with wrong data.
Concerning the utilizing pivot point, why not? you may feed your net with pivot point values as inputs and future percent change of close/open as output. I would like to try them also.
As far as I know Noxa has limitation, i.e. it should be adjusted manually in order to matched with the market cycle (peak and valley of price data), as long as the frequency is matched that you will get profit from it. I received a statement from someone that it is difficult to use GA automatically to find the best noxa parameter. If you feed them to the net with different frequency from the market, then you feed your net with irrelevant data, you can not expect it will produce profitable trade.
Personally, I made some investigation to find what are the best output that can be used for my net, which inputs shall be selected, how, etc. I believe Neuroshell has capabilities to verify them. I prefer to return to basic/core idea how we can utilize NN, instead of trying all indicators without knowing what we dealt with (one of my previous mistake).
Arryex
My most successful neural nets
I have a good system that resembles Mr Future bi-distribution I think. It does not involve neural nets:
First I set a cssa cycle manually and then extract two instantaneous periods from it; one from the tops and one from the bottoms. Next, I get both distributions from a sliding window. You’ll notice that the distributions are skewed differently whether the market is in an up or a down trend. I finally set my entries from these distributions accordingly. Results are quite good and robust.
Mr Future, your comments?
Amit,
My most successful neural nets use cssa cycles but I don’t predict levels. Rather I train my net predicting the current close of a ssa cycle and look whether the composite produced leads or lags the cycle. All cycles are set manually. I trade only instruments which cycle composite leads the SSA cycle. It’s very much like a switch telling me what instrument and what cycle to trade. These systems work very well for more than a year now.
Tuen
I have a good system that resembles Mr Future bi-distribution I think. It does not involve neural nets:
First I set a cssa cycle manually and then extract two instantaneous periods from it; one from the tops and one from the bottoms. Next, I get both distributions from a sliding window. You’ll notice that the distributions are skewed differently whether the market is in an up or a down trend. I finally set my entries from these distributions accordingly. Results are quite good and robust.
Mr Future, your comments?
Amit,
My most successful neural nets use cssa cycles but I don’t predict levels. Rather I train my net predicting the current close of a ssa cycle and look whether the composite produced leads or lags the cycle. All cycles are set manually. I trade only instruments which cycle composite leads the SSA cycle. It’s very much like a switch telling me what instrument and what cycle to trade. These systems work very well for more than a year now.
Tuen
Re: My most successful neural nets
Hi Nevis,
Thanks for your explanation, your statement still making me to confirm that CSSA should be set up manually. Some time it will not follow the market change and the parameter to be readjusted. So far there is no auto adjustment/adaptive features, and it is difficult to use NN or GA to find the proper CSSA parameters.
Please advice if you can state it is a robust system if its not an adaptive to a market change.
Thanks,
Arryex
Hi Nevis,
Thanks for your explanation, your statement still making me to confirm that CSSA should be set up manually. Some time it will not follow the market change and the parameter to be readjusted. So far there is no auto adjustment/adaptive features, and it is difficult to use NN or GA to find the proper CSSA parameters.
Please advice if you can state it is a robust system if its not an adaptive to a market change.
Thanks,
Arryex
I’m hoping that someone here can help me… I just got and installed Noxa Causal Singular Spectrum Analysis (CSSA). For some reason it crashes my NeuroShell! It becomes unresponsive and I need to force close it! So I started thinking… I would install it on another computer to test if it worked there, and in fact it does! I just really want it on my laptop instead! Has anyone had this problem? And if so… how did you fix it? I thought that there might be some issue with my laptop so I just wiped it and reinstalled everything… still the problem is there!
I would greatly appreciate any help you can offer!
I would greatly appreciate any help you can offer!