Trading Strategies Based On Digital Filters - page 59
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
//We check for amplitude peaks and tag them
for(i=MinPer+1; i<MaxPer; i++)
{
if(AmpBuf>AmpBuf && AmpBuf>AmpBuf)
PicBuf=i*MathPow(10,-1*digs);
else
PicBuf=0.0;
Pics are found of spectrum amplitudes and tagged. Later on you can just compare the tagged amplitude pics to threshold value and sort.
I believe threshold value should not be fixed inside the code but as a external
changable parameter so optimizer can find optimal value.
Krzysztof
Peace in the Valley
Really good to see peace here!!!!! After all it is about puttinng pips into the account. You both are clearly smart people better to work in the same direction rather then using energy to argue. I love the cleaness of your charts Simba!
Good Trading
Jim
Really good to see peace here!!!!! After all it is about puttinng pips into the account. You both are clearly smart people better to work in the same direction rather then using energy to argue. I love the cleaness of your charts Simba!
Good Trading
JimHahahaha,Thanks Jim,long time no see...I rent my charts at "rentaclean chartdotcom" they are expensive but simplify the issue of trading
How is everything going?Any comments on cycles?
Regards
Simba
PLEASE , somebody answer to my cuestions in post 518.
Thank you for help.
Excuse me for my poor English.
First of all , lets somebody tell me which program is best for Spectral Analysis?
I make it with free program"Digital Filters Methods" ,
then , i try pay program "Spectral Analyzer" see pictures bellow.
How you see analisys is different and i dont know who is right?
Please see differrents on 2 pictures make it with "Spectral Analyzer" , i only change correlation matrix button (mark with black pen) , and view how is different anallysis.
So , lets people with more experience let me know how to make spectral analisys and which program is best.
And seccond , when we see spectral analysys , how we have to make "perfect cycle indicator" with free program"Digital Filters Methods", How to understand who is large cycle?
Who is best settings to P1 , D1 , P2 , D2?
I try to make this indicator with Simba method on post 253.
I put for P1-74 , P2-76 (to isolate Peak 75) and D1-57 , D2-96(bottoms) , so a get some indicator , when i change D1 and D2 , with little differents nummbers D1-56 , and D2-97 , i make absolutelly different indicator , so i thing this is not right method to make "perfect cycle indicator"
Thank you for help.//We check for amplitude peaks and tag them
for(i=MinPer+1; i<MaxPer; i++)
{
if(AmpBuf>AmpBuf && AmpBuf>AmpBuf)
PicBuf=i*MathPow(10,-1*digs);
else
PicBuf=0.0;
Pics are found of spectrum amplitudes and tagged. Later on you can just compare the tagged amplitude pics to threshold value and sort.
I believe threshold value should not be fixed inside the code but as a external
changable parameter so optimizer can find optimal value.
KrzysztofThreshold value should be as wide as necessary...as per the tests...BTW ,Don`t buy GBPJPY in the next couple of days
...Look for selling spots...this is real advice(as always,do it on demo 
),good or bad,reality will say
Don`t mistake Hattori Hanzo with Black Mamba...
PLEASE , somebody answer to my cuestions in post 518.
Thank you for help.
Hi,
Use the latest one...D1,P1,P2,D2 As 68,80,161,183..in case strange unfit cycle happens,expand D1 and D2(68 and 183..to 67(66,65...) and 184(185,186...))until you have a fit.
The 2 peaks at 80 and 161 are probably harmonics.
Yes,you were right in your pm,I hadn`t understood your question,hope this now solves your problem,and ,if possible ,be so kind to post pictures of your cycle applied to the chart.
EDIT:Try,additionally, 60,80,92,161..and see what happens
Regards
Simba
no causal SSA
Threshold value should be as wide as necessary...as per the tests...BTW ,Don`t buy GBPJPY in the next couple of days
First I also think there is enough money on Forex for everybody here
Regarding this picture. I think you are using non causal SSA in this. Than are you sure that that repaint of buy/sell signal is not occuring here ??
Example of such repaint is here
Neural Network Trading, Serious people only! - Page 6
post 88 and down.
That's the impact of non causal indicator in the strategy, if you look to the screenshoots you see that buy/sell signal are adjusting them selves.
It's very hard to dedect this thing, for sure not on 4H TF, only on 1m or 5m TF by observation. All statistics of trading strategy are faked with too good results.
Here is the piece of CSSA help
Krzysztof
Gbpjpy down
YES SIMBA VERY GOOD ADVISE, LOOKS LIKE STARTING TO GO DOWN
,GOOD CALL!
tools
First I also think there is enough money on Forex for everybody here
Regarding this picture. I think you are using non causal SSA in this. Than are you sure that that repaint of buy/sell signal is not occuring here ??
Example of such repaint is here
Neural Network Trading, Serious people only! - Page 6
post 88 and down.
That's the impact of non causal indicator in the strategy, if you look to the screenshoots you see that buy/sell signal are adjusting them selves.
It's very hard to dedect this thing, for sure not on 4H TF, only on 1m or 5m TF by observation. All statistics of trading strategy are faked with too good results.
Here is the piece of CSSA help
KrzysztofInternational Journal of Forecasting 25 (2009) 103–118
Abstract
In this paper, the performance of the Singular Spectrum Analysis (SSA) technique is assessed by applying it to 24 series
measuring the monthly seasonally unadjusted industrial production for important sectors of the German, French and UK
economies. The results are compared with those obtained using the Holt–Winters’ and ARIMA models. All three methods
perform similarly in short-term forecasting and in predicting the direction of change (DC). However, at longer horizons, SSA
significantly outperforms the ARIMA and Holt–Winters’ methods.
END OF ABSTRACT
BTW..Long term horizon was >=3 months(3 Bars,since they used Monthly data).
My opinion:
SSA is especially useful for analyzing and
forecasting series with seasonal components
and non-stationarity...we could discuss the seasonality degree in Forex data,not the non stationarity ,IF Forex were stationary there would be no game,no counterparts to your trades,too easy to predict....I will summarize the actual fashionable methodology among econometricians for using SSA...for the sake of readers.
Fisrt:Use SSA to decompose the
original series into a sum of a small number of subseries, so that each subseries can be identified as
either a trend,a periodic/quasi-periodic Cycle
or NOISE...This is
followed by a reconstruction of the original series...obviously without noise
...How?
Next:You use orthogonality(lack of correlation) and distancebetween eigenvalues to determine how many of them to use to reconstruct the original signal(which is discrete,by the way)...and WHICH ONES YOU DISCARD ...ok,so we take the different "uncorrelated factors" until the point where they start "grouping",the rest we discard.
Next: Simulate several hundreds or a few thousands independent
copies of the process and apply the forecasting procedure to those independent time series ....In this way a Monte Carlo average probability can be guesstimated,BTW,I note that this process implies a belief in a gaussian PDF..which doesn`t happen in Financial time series ,so,the high probability is just a guesstimate...Vivan los nerds.
Next:Compare actual Forecast with "average" Montecarlo(boostrap)..if too far from it discard...if not...
Next:Compute confidence intervals..again,biased by incorrect PDF Assumption
So,even if SSA is loved by econometricians,it just provides them a "less bad "guesstimate,for the moment
...BUT... SSA is extremely useful in filtering out noise without creating distortions(ok,creating minimum distortions) in the trend and cyclic qualities(if any) of the original time series.
BTW,HP filter(which is also loved by econometricians) just does a very similar job,and is less computer intensive...In any case you can`t use either of them,like you can`t use an EMA, for predicting,just to denoise .
Causal,non causal...why this separation?..prices evolve,so both sets of tools are useful...you can use either...again,it is not the tool,it is the artist that matters..and an artist is just a clever and interested person with 10k hours practice with the tools..You think this is science,and,no,this is an art,and,obviously,like in every art(think jazz music,a jam session with no partiture...)the practitioner needs a minimum mastery of the scientific elements of the art(rhytm,melody,composition,etc),but the key element,like in a jam session,is the hability to adapt ,fasjt...Yes,I know you are thinking about it
Why don`t you post some "predictions" using Noxa CSSA for actual Forex pairs?...This will be very interesting,much more than trading "CHIRPS",nobody is gonna point a finger to you for failing..only for not trying.
Ciao
Simba
International Journal of Forecasting 25 (2009) 103–118
Abstract
In this paper, the performance of the Singular Spectrum Analysis (SSA) technique is assessed by applying it to 24 series
measuring the monthly seasonally unadjusted industrial production for important sectors of the German, French and UK
economies. The results are compared with those obtained using the Holt–Winters’ and ARIMA models. All three methods
perform similarly in short-term forecasting and in predicting the direction of change (DC). However, at longer horizons, SSA
significantly outperforms the ARIMA and Holt–Winters’ methods.
END OF ABSTRACT
BTW..Long term horizon was >=3 months(3 Bars,since they used Monthly data).
My opinion:
SSA is especially useful for analyzing and
forecasting series with seasonal components
and non-stationarity...we could discuss the seasonality degree in Forex data,not the non stationarity ,IF Forex were stationary there would be no game,no counterparts to your trades,too easy to predict....I will summarize the actual fashionable methodology among econometricians for using SSA...for the sake of readers.
Fisrt:Use SSA to decompose the
original series into a sum of a small number of subseries, so that each subseries can be identified as
either a trend,a periodic/quasi-periodic Cycle
or NOISE...This is
followed by a reconstruction of the original series...obviously without noise
...How?
Next:You use orthogonality(lack of correlation) and distancebetween eigenvalues to determine how many of them to use to reconstruct the original signal(which is discrete,by the way)...and WHICH ONES YOU DISCARD ...ok,so we take the different "uncorrelated factors" until the point where they start "grouping",the rest we discard.
Next: Simulate several hundreds or a few thousands independent
copies of the process and apply the forecasting procedure to those independent time series ....In this way a Monte Carlo average probability can be guesstimated,BTW,I note that this process implies a belief in a gaussian PDF..which doesn`t happen in Financial time series ,so,the high probability is just a guesstimate...Vivan los nerds.
Next:Compare actual Forecast with "average" Montecarlo(boostrap)..if too far from it discard...if not...
Next:Compute confidence intervals..again,biased by incorrect PDF Assumption
So,even if SSA is loved by econometricians,it just provides them a "less bad "guesstimate,for the moment
...BUT... SSA is extremely useful in filtering out noise without creating distortions(ok,creating minimum distortions) in the trend and cyclic qualities(if any) of the original time series.
BTW,HP filter(which is also loved by econometricians) just does a very similar job,and is less computer intensive...In any case you can`t use either of them,like you can`t use an EMA, for predicting,just to denoise .
Causal,non causal...why this separation?..prices evolve,so both sets of tools are useful...you can use either...again,it is not the tool,it is the artist that matters..and an artist is just a clever and interested person with 10k hours practice with the tools..You think this is science,and,no,this is an art,and,obviously,like in every art(think jazz music,a jam session with no partiture...)the practitioner needs a minimum mastery of the scientific elements of the art(rhytm,melody,composition,etc),but the key element,like in a jam session,is the hability to adapt ,fasjt...Yes,I know you are thinking about it
Why don`t you post some "predictions" using Noxa CSSA for actual Forex pairs?...This will be very interesting,much more than trading "CHIRPS",nobody is gonna point a finger to you for failing..only for not trying.
Ciao
SimbaI made already out of sample test for EURUSD for 1,5 and 15m TF for NOXA CSSA on TRADE2WIN so you can have a look.
I know that SSA is used for denoising but the reaction of trading sytems will be repaint and faking buy/sell signals and strategy results.
So what is a conclusion ??? Was this picture form yesterday and results of yours strategies faked by SSA repaint or not ?? Than how the real picture looks like ??
The example of repaint which I posted was giving 80% hit but when I removed non causulal element it went down to 42%.
Krzysztof