Trading Strategies Based On Digital Filters - page 19
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
1.
>>No,when working with digital filters I don`t prepare quotations for analysis.
attachment eur d1 1 pic.- 2000 bar 2pic - 1000 bar what is the meaning of the attachment,could you explain ?[/B]
>>I tried once to apply a standard SATL to a very short Fatl(smoothed real time representations of price),and the only thing I achieved was the cpu jumping at 98 degrees Celsius,just before the system stopped working.
Do you use fatl with dll ??? no ,I applied the code to the code,just a trial
>>3-One important thing I do when doing digital filters is trying to find 2 harmonic cycles in 2 harmonic timeframes,i.e:the representation of the same cycle in both timeframes..M30 you have a dominant period of 32..then at M15 you have a dominant period of 64..this cycle is going to be,first:very useful at most timeframes,second:very long living(high bartels).
Did you try use for analysis non-standard timeframe? D3, H9 etc. ? NOdo you analysis alpari quotations or other broker?for historical analysis I used alpari,now I use whatever data I might have,all brokers data is valid for short term periods,like 200,where none is good for longer,I used to do analysis with cycletrends and EOD data that I bought to a very good data provider,frankly,data is much better,results are similar,so,I prefer to do it simpleplease find my reply above
dvarrin..any reply?
Sounds great. Dvarrin, please let us know if that response helped. I want you to be up to speed and comfortable before we start talking about other things. cl
Hi clahn04,
Thank you very much for your answer! I'm sorry for the late reply. I was quite busy during those last few days :-(
What you wrote corresponds to what Simba explained us too :-) It was on my side that I was a bit lost, because I was looking for an indicator that could show the market cycles and I mixed a bit everything up with the number of coefficients to use
About the STLM, I modified an existing one with the new coefficients and it's working fine too. So I think we can go on with RBCI :-):-)
About your strategy, is it the one you explained me already (buy when STLM is crossing 0) ?
I'm looking for something a bit more efficient than Hurst cycle's timing and I thought that STLM could help me, but the problem is that if it is only a delayed RSTL, it is showing what the last wave did, but not what the current wave is likely to do and when it is going to reverse.
Thanks a lot again!!
Cheers,
No...it's a much different strategy....and for now it's more just an idea...i am ready for RBCI too........
I can't seem to calculate RBCI with that software correctly....at least i don't think it's correct....RBCI is simbly FATL - SATL, correct SIMBA?
I actually just cheated and took a STLM histogram, replaced the coefficients with FATL and SATL, and made a homemade RBCI and it works fine....i had to make do for the time being...
In any event, i'm ready to create whenever you guys are.
cl
Before we move on...i guess i have a house keeping question...
Simba you mentioned markets being fractal.....should we always use a certain time frame for our spectral analysis?...and then apply those values to all time frames.....or, should we do something like 4H can be used for 1H, 4H, an Day, and 15 min can be 5, 15, 30.....
I think that would help me to add some consistency if i knew a "best practice"
cl
SIMBA:
>>what is the meaning of the attachment,could you explain ?
The result depends on number of sample elemets. It hapend because we have not stationary time series and not Gaussian process. We need to determin time series part were time series a clouse to stationary and made analysis for this part.
SIMBA:
>>what is the meaning of the attachment,could you explain ?
The result depends on number of sample elemets. It hapend because we have not stationary time series and not Gaussian process. We need to determin time series part were time series a clouse to stationary and made analysis for this part.And how do you determine the part of the time series that is (WAS
)semi stationary ?i have used bartels test with cycletrends software,have used wavelets with 20 years data,clean data, on D1 and ,frankly,in real life,it doesn`t give a better result than just using all available history(like 7 years of m1 data,for example)in Alpari and choosing the highest peaks.
What I have found is that by reoptimizing every week for the last 200 periods,for h4 and h1,you have very good probabilities to nail it for the next week...anyway ,I don`t pretend to know any truth,I just focus on the practical,so,if you want to post some examples I will be glad to learn.
dvarrin:glad to know you are back with us.Using custom stlm2 zero cross will not work,it is the same as using satl,rstl cross,and ,for the custom satl and rstl we did,with 16 and 22/24 coefficients,it doesn`t work..what worked was the rstl slope for entries and exits and reverses,On Monday or Tuesday I will be back at my usual location and will be able to post the EA,so that you can check..then we can substitute the custom indicators with new ones like RBCI
or new optimized SATL,RSTL,STLM2...and test our ideas.
clahn04:Do you care to explain about your strategy?
both:Could you post the SATL,RSTL and stml /rbci that you have created,if any?
Thanks..and have a good weekend
Simba
No...it's a much different strategy....and for now it's more just an idea...i am ready for RBCI too........
I can't seem to calculate RBCI with that software correctly....at least i don't think it's correct....RBCI is simbly FATL - SATL, correct SIMBA?Did you check the code?
I actually just cheated and took a STLM histogram, replaced the coefficients with FATL and SATL, and made a homemade RBCI and it works fine....i had to make do for the time being...Did you replace them with a custom SATL and custom FATL,or just the standard ones?
In any event, i'm ready to create whenever you guys are.
clHi clahn04,please post your custom RBCI,if you don`t mind..or just try to do one ,if you have free time
,by using d1=14,d2=115...and use 2 intermediate peaks as P1 and P2..You will obtain something very similar to a cycle,if you do it once the market is closed(use 200 to 250 periods..whatever suits you better,and apply the spectrum analyzer to them)we can use it to test for next week .
Thanks and Regards
Simba
SIMBA:
>>And how do you determine the part of the time series that is (WAS )semi >>stationary ?
I use finware software. I can provide it with admin permission.
>> i have used bartels test with cycletrends software
Can you provide me with link (pm please)
hope this isn't too long as 1 post
Hi everyone,
I want to join the fun and learn how to do this, so that I too can use Simba's EA when it gets posted. I want to make sure that I am doing this correctly before I make differents sets to compare one set of P1/D1 numbers vs another set.
In my case I'm using IBFXm GBPUSD, 1HR data. The reason why I did not use 4HR at this point is because the 4HR bar looks different for different broker times. I'm not sure which brokers Simba, clahn04, and dvarrin are using...so this way it might be easier to check my work when placing it with your charts.
1) I opened & analyzed the data through DFM's Spectral Analysis. I chose the peaks 41 & 12 (see screenshot below).
2) Using the DFM's Filter Generator, changed P1=41 & D1=12 to generate my SATL. By using these 2 boundaries, all the cycles in between (15,21,30) get smoothed out in the SATL indie..yes/no?
3) To generate my RSTL I used the same Filter Generator, I kept P1 & D1 the same numbers as the SATL and then changed "Delay, bar" to half the number of coefficients in the SATL generated formula.
Am I understanding correctly, the coefficient is the following?
The part of the formula that says...
response=
0.2134009687843*Close
+0.2038536296690*Close
+0.1859934562656*Close
...
-0.01037729654480*Close;
Half of 17 is 8.5
4) I generated a few RSTL to see what they would look like. The 3 "Delay, bar(s)" that I used were 7, 8, & 9. If my understanding of coefficient is correct, then RSTL_41_12,delay7 had 23 coefficients, RSTL_41_12,delay8: 24 coefficients and RSTL_41_12,delay9: 25 coefficients.
This falls within the range of 1.5 times the coefficient of SATL's 17, which is approximately 25.5
A screenshot of the indie's I created are below. SATL is Magenta, RSTL_delay7:Gold, RSTL_8:Red, RSTL_9:FireBrick. The same color scheme matches the STML.
5) I created the STML2 indie by using one that was posted by NewDigital as a guideline. I then followed clahn04's cut&paste suggestion. Could someone please check to make sure I did this section correctly?
Inside the STML2 indicator there is a formula that reads
STLM=value1-value2;
STLM1=value3-value4;
From what I could tell value1 & value3 is represented by SATL based upon the equation for STLM, "STLM(k)=SATL(k)–RSTL(k)". Value2 & value4 represents RSTL.
I changed the formulas for value1, value2, value3, and value4 with the respective coefficients found in the SATL & RSTL indies. I made sure to keep the Close[xyz] the same as the Original STML.
value1 =
+0....*Close
+0....*Close...;
value2 =
+0....*Close
+0....*Close
...;
value3 =
+0....*Close
+0....*Close
...;
value4 =
+0....*Close
+0....*Close
...;
The STML was pretty easy to generate with a couple of copy/paste and concatentate steps in Excel. If someone can confirm that I did the above steps correctly, I'd be more than happy to share the excel template I made and add instructions to it.
QUESTIONS:
1) I really don't see that much difference between the STLMs created with the different delay bars. In the snapshot of the H4 chart, there are no differences. In the H1 chart, there are 2. DelayBar=9, is slower. DelayBar7 and DelayBar8 look to be the same. Does this mean I should step down to DelayBar6 to see if it speeds up the signal? Am I ok to keep stepping down the DelayBar to get a faster reaction as long as the # of STML's coefficients is within 10% of 1.5*SATL's coefficients?
2) Broad statement/question. I'm still trying to grasp the whole Spectral Analysis thing. Why do I use 1 number over another for our SATL inputs? Just because it's a peak?
3) In the Spectral Analysis I've posted, should I consider using 57 over 41? 41 is higher by 0.02. What does the y-axis represent?
4) What is the Lowest or Highest Frequency (x-axis) that I should consider using for my SATL inputs? In other words, how far left (shorter frequency) or far right (longer frequency) should I consider? I guess the answer would be dependent upon the timeframe being analyzed. What are good "boundaries" for H1 & H4?
5) Although the frequency spike is higher at 12 than it is at 21, should I have considered using 21 as D1 because that is half the frequency of the tallest spike of 41 (P1)?
6) How good are the results from the Spectral Analysis? The author of the software wrote this in the Help Section
The chosen spectrum analyzing method is not the best. The author will be thankful for any offer regarding its improvement.
OK, I'm ready for any feedback and am semi ready for the next step! Have a great weekend!