Trading Strategies Based On Digital Filters - page 47
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
detrending and denoising
I believe this paper can make purpouse of detrending easier to understand.
Regarding denoising. I think purpouse is to extract a 'clean cycles' and measure
S/N ratio also as when S/N ratio is high then are higher chances to make sucessfull trades and for this we have to measure the 'noise'
S/N ratio is also well described in Codebreaker thread.
Krzysztof
I believe this paper can make purpouse of detrending easier to understand.
Regarding denoising. I think purpouse is to extract a 'clean cycles' and measure
S/N ratio also as when S/N ratio is high then are higher chances to make sucessfull trades and for this we have to measure the 'noise'
S/N ratio is also well described in Codebreaker thread.
KrzysztofI read the paper. It is a good one, but I am not sure that detrending and denoising (that is "filtering") is a good choice as a preprocessing for an analyzer.
As you denoise and detrend, you are modifying the spectrum, so I guess it is better to analyze the original signal, maybe with different windows or even with different analyzers, and find out if cycles are recognized in each one of them.
That's really great!! :-)) It works fine with the 40 and 44 :-)
How do we choose the valleys for D1 and D2? Should they be as low as possible? if there is a peak next to the peak we will use to define P1 and P2, do we have to take the valey between them?
Do you have some example on which values we could use for method 2?
How do we choose the two peaks inside and the 2 peaks on the order. Are there some ideal configurations?Hi dvarrin,
I attach an example. In the first subwindow there is the barchart and the filtered signal. The signal is filtered with the filter I posted yesterday. The values are chosen from MESA analysis over a short window of 200 bars, with an autoregression order of 150. You can see the spectrum shape in the second subwindow and the values of peaks and valleys in the third subwindow.
I have chosen for the filter the values P1=70 and D1=52.
Bye
pdf
is here - Meyers analytics
detrending
Hi,
Please look at this document. Here is an key point page 2
This is not too helpful! What happened? This is an example of what happens when the trend and series average are not taken out of the time series before the FFT is done. The trend and the average completely swamp the frequency domain such that none of the characteristics we are
looking for can be found.
in simple words. We have f=x+sinx, we are interested in sinx only and we don't need x
Krzysztof
NOXA CSSA cycle EURUSD30
Here is a cycle which i found using NOXA CSSA for 30 min EURUSD. Seems to be very profitable > 400 pips over two days out of sample, I will keep those settings and we will see if it will earn money over tomorrow - i doubt.
You can consider this cycle as a combined cycle, line in the upper chart - pure trend without cyclical components. Lower window cycle for short (red) and long (blue) entries found using genetic optimizer.
The cycle is denoised and detrended of course (theoretically)
So you can compare your cycles with this one.
Krzysztof
Digital filters really usefull ???
Hi,
I've been playing with digital filters creation for sometime, but I'm wondering if they are really of any help to improve the trade entries compared to using some Moving Averages.
I've created the SATL, RSTL and STLM filters for EURUSD 30M
I could find the peaks to be at 18, 42 and 79.
Then, using a simple MACD indicator, I'm setting the short EMA to 42/2 = 21, the long EMA to 79/2 = 40 and the signal EMA to 18/2 = 9. Setting short EMA to 42 and Long EMA to 79 is also possible.
Look at the chart below. The MACD is crossing the signal line at about the same time when SATL is crossing RSTL.
frequency domain
You can only compare them in frequency domain but i think it is not a problem at all. DFG has a possibility to compare filters, and accepts .ex4 files than put MACD there and compare impuls response. Or you can generate time series with MACD and clone as a filter and compare. It is described in help.
Otherwise it can by simple luck that it match.
Krzysztof
are they usefull ??
are they usefull ??
Thats the problem with this thread, it lasted ages but it was never properly concluded with an answer of this question.
Anyway if you want just give me parameters and tell what are the buy/sell signals and I will make a strategy with NS and we will see. We can compare also to standard MACD strategy.
I was thinking about doing this but never had time.
you can prepare the parameters for example based on data till last friday, so Monday and Tuesday and today will be out of sample test. Than I can find optimal parameters of filters and we will see what was set wrong.
Krzysztof
Hi,
I've been playing with digital filters creation for sometime, but I'm wondering if they are really of any help to improve the trade entries compared to using some Moving Averages.
I've created the SATL, RSTL and STLM filters for EURUSD 30M
I could find the peaks to be at 18, 42 and 79.
Then, using a simple MACD indicator, I'm setting the short EMA to 42/2 = 21, the long EMA to 79/2 = 40 and the signal EMA to 18/2 = 9. Setting short EMA to 42 and Long EMA to 79 is also possible.
Look at the chart below. The MACD is crossing the signal line at about the same time when SATL is crossing RSTL.Dvarrin,
A few questions.
1. How did you create your RSTL? That doesn't look like any RSTL i have ever created.
2. Trading the turn of STLM isn't the correct way to use the digital filters IMHO. STLM+SATL defines the "direction" that you trade in.
There are just a lot of "rules" with regards to these particular filters that are in the ATCF document in this thread that I think are worth reading. No matter what strategy you trade, I just think it's important to know the intentions of certain indicators (see my comment about CCI).
cl