Trading Strategies Based On Digital Filters - page 58
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
I love how each post you accuse me of one more thing, but yet each post you never defend any of my accusations?
You're almost as jittery as a person who's gone cold turkey on crack.
And no, I am not a part of that "secret" group, but I am actually a part of several other "secret" groups which are "secret" because we don't personally invite people like you. =)
Wow !!! I just compared results of three trading systems plus uncovered 'secret' group which was hunting for 'free' researchers and programmers and a lot of noise !!!!! Maybe median filter will help
Anyway still nothing about digital filters from your side....are you member of this group also ???
KrzysztofHi dvarrin,
the analyzer currently posted (R-MESA_instant_spectrum.mq4) has an option to denoise ("FilterMode" parameter) that I made on the fly to see how different types of filters behave as denoiser
0: no denoising
1: Movin average
2: MedianFilter - simple non linear filter (mine)
3: NonLagMA_v6.1 - you can find it anywere
4: Kalman - free kalman implementation for MT
5: JMA - same as above (don't know if it's reliable)
6: R-DigitalFilter - this is mine, you can find it here
Note that you must have filters from 2 to 6 installed as custom filters in your MT. If you have not, you will get crazy results.
Because of my priorities, I have not integrated denoising in "R-MESA-Cutoff_frequencies.mq4" indicator yet (although is trivial), so all the others depending on the latter are without denoising.
Please be aware that the "smoothed" parameters in this last one does not refer to a denoising preprocess, but to a post-process smoothing of mesa peaks through sliding windows (so that you don't have to change filter every now and then).Sorry to be so noob richcap, but what would those denoiser exactly do on the price? What is the advantage of using them? Does it means that instead of using the price data, it will base the spectrum analysis on one of those denoisers?
I love how each post you accuse me of one more thing, but yet each post you never defend any of my accusations?
You're almost as jittery as a person who's gone cold turkey on crack.
And no, I am not a part of that "secret" group, but I am actually a part of several other "secret" groups which are "secret" because we don't personally invite people like you. =)By commercial i meant if you sell EAs or indicators.
I see from your blog that you are not 'full timer', some side job and were complaining that I'm full timer just 2 years ??
So do you have MBA like Simba ?? The same school ??
Krzysztof
Hi Richcap,
can you have a look to this
T2W Day Trading & Forex Forums
and comment to NOXA if conditions in your tests are fullfilling out of sample conditions and we don't have a backward curve fitting here.
KrzysztofKrzysztof,
I agree with the guys at NOXA that you compared apples with oranges.
You asked if I was able to make money from your unrealistic noisy chirp signal and it was fun for me to show you a 100% winner - infinite profit factor strategy.
I have already explained how I did it, so it should be clear that I changed the default setting of max_period parameter from 140 to 800, otherwise my indicator that makes use of MESA to find peakes in the spectral density power, would never recognize the peaks over 200 that your signal have. I have then applied a very simple strategy on this indicator.
If I was to make a commercial product, I would easily integrate a wideband peak scanner as a first step to set properly (and automatically) that parameter of the adaptive indicators.
I agree with you that systems must be verified with scientific method and I'm trying to do it (with your help too) for the Spectral Analisys. In that process, your chirp signal were useful.
But from a verified spectral analyzer to a complete trading strategy there's a long way to go. And it seems that you don't have the patience to help (or at least wait) for that way to be walked.
Am I wrong?
Krzysztof,
I agree with the guys at NOXA that you compared apples with oranges.
You asked if I was able to make money from your unrealistic noisy chirp signal and it was fun for me to show you a 100% winner - infinite profit factor strategy.
I have already explained how I did it, so it should be clear that I changed the default setting of max_period parameter from 140 to 800, otherwise my indicator that makes use of MESA to find peakes in the spectral density power, would never recognize the peaks over 200 that your signal have. I have then applied a very simple strategy on this indicator.
If I was to make a commercial product, I would easily integrate a wideband peak scanner as a first step to set properly (and automatically) that parameter of the adaptive indicators.
I agree with you that systems must be verified with scientific method and I'm trying to do it (with your help too) for the Spectral Analisys. In that process, your chirp signal were useful.
But from a verified spectral analyzer to a complete trading strategy there's a long way to go. And it seems that you don't have the patience to help (or at least wait) for that way to be walked.
Am I wrong?OK, I wasn't sure how you did this this is a reason that I wrote methodology
post to double check but you haven't replied somehow.
So does the same applies to the second test ??
Krzysztof
OK, I wasn't sure how you did this this is a reason that I wrote methodology
post to double check but you haven't replied somehow.
So does the same applies to the second test ??
KrzysztofKrzysztof,
it is my conviction that there is only one way to do things right: to understand what you are doing.
Did you understand what you were doing with your silly challenge?
If not, you were wasting your time (not mine, because I have kept on learning from my actions).
If yes, you were only trying to gain a personal advantage from this challenge (i.e. having disclosed the inner knowledge of someone's hard work, no matter if they do it for money, for passion or for glory).
I'm sorry I took it as a joke at first. There is no joke in mocking other's work.
There are a lot of "engineering" ways to model the market. I myself have an endless todo list. And it's not a list like this "try this indicator, try this software, try this strategy and, eventually, if I'm sure at 100% that it works, buy it".
Why don't you chose one and show that you are able not only to do your past job of dissecting other's work, but even to contribute with fresh ideas and findings?
Sorry to be so noob richcap, but what would those denoiser exactly do on the price? What is the advantage of using them? Does it means that instead of using the price data, it will base the spectrum analysis on one of those denoisers?
Yes, if you chose a FilterMode other than 0, raw data are filtered before passing them to the spectrum analyzer.
Basically, you modify the spectrum of the raw data to make life easy to MESA spectrum analyzer
.
Moving average (SMA) is the most known filter. What DSP oriented trader don't like in MA is that they give you no control over the spectrum distortion they introduce on signal (not to talk about phase lag).
Median filter was an experiment. As noise in market timeseries is non linear, I decided to put one of the the easyest non linear filters on the bench.
The other's are well known causal and non causal filters. I have not implemented SSA because I had no code to rework at the moment and I don't know SSA theory and algorithm.
Every body with an SSA filter available can easily add it to the indicator (if he knows a bit of 'C' or 'mql' and of metatrader platform)
Yes, if you chose a FilterMode other than 0, raw data are filtered before passing them to the spectrum analyzer.
Basically, you modify the spectrum of the raw data to make life easy to MESA spectrum analyzer
.
Moving average (SMA) is the most known filter. What DSP oriented trader don't like in MA is that they give you no control over the spectrum distortion they introduce on signal (not to talk about phase lag).
Median filter was an experiment. As noise in market timeseries is non linear, I decided to put one of the the easyest non linear filters on the bench.
The other's are well known causal and non causal filters. I have not implemented SSA because I had no code to rework at the moment and I don't know SSA theory and algorithm.
Every body with an SSA filter available can easily add it to the indicator (if he knows a bit of 'C' or 'mql' and of metatrader platform)Thank you very much for your answer and patience Richcap!
Krzysztof,
it is my conviction that there is only one way to do things right: to understand what you are doing.
Did you understand what you were doing with your silly challenge?
If not, you were wasting your time (not mine, because I have kept on learning from my actions).
If yes, you were only trying to gain a personal advantage from this challenge (i.e. having disclosed the inner knowledge of someone's hard work, no matter if they do it for money, for passion or for glory).
I'm sorry I took it as a joke at first. There is no joke in mocking other's work.
There are a lot of "engineering" ways to model the market. I myself have an endless todo list. And it's not a list like this "try this indicator, try this software, try this strategy and, eventually, if I'm sure at 100% that it works, buy it".
Why don't you chose one and show that you are able not only to do your past job of dissecting other's work, but even to contribute with fresh ideas and findings?Well, you simple provided certain results and it seems that they can not be used in real trading as so far your system is semi manual and simple backward curve fitting at least in terms of results which you provided. You didn't explain it clearly to the readers...
Than of course we can not compare apples to oranges....
I hope you didn't get upset that i involved NOXA designer as you could see from his answer that he is very competent person and we can only gain
brain storming with him.
Regarding my impact. I m working all the time on it, writing dll for Goertzel to evaluate it properly, i will also post simple piece of code of missing 'wideband scanner' from you system than everybody can add this. This piece is included in Goertzel V1 indicator already so everybody can have a look.
I don't know if you will like this piece or not but for sure it can change results
dramatically.
Regarding disclosing etc. I didn't disclose anything here, it was a common knowledge.
Anyway I will answer to NOXA designer with clarification.
Krzysztof
Regarding my impact. I m working all the time on it, writing dll for Goertzel to evaluate it properly, i will also post simple piece of code of missing 'wideband scanner' from you system than everybody can add this. This piece is included in Goertzel V1 indicator already so everybody can have a look.
I don't know if you will like this piece or not but for sure it can change results
dramatically.
Fine, I'm waiting