Trading Strategies Based On Digital Filters - page 13

[dvarrin]

SIMBA:
Hi Daniel,my reply is above,why don`t you just do the SATL and PCCI and post them here so we can take a look and see if they need some modifications?I explained to you one of the ways to do what you requested,usually it is the most useful one,but there are alternatives in case the end result doesn`t models the price as it should.

Regards

Simba

Hello Simba,

Thank you very much for your reply!! I'll create a SATL and a PCCI indicator and post them so that you can check if they are looking good.

If I want to create the SATL and FATL indicators, I think I've to use the short periods for FATL and the long periods for SATL. Am I right? Is it ok for example to take P1=73 and D1=30 for SATL and P1=12 and D1=21 for SATL?

And in this case, I should not divide the values by 2, like for a normal stochastic or RSI indicator?

About the filters' sets, how do we combine the filters if I want to create a STLM - FTLM indicator for example?

Regards,

Daniel

[SIMBA]

filters

dvarrin:
Hello Simba,

Thank you very much for your reply!! I'll create a SATL and a PCCI indicator and post them so that you can check if they are looking good.OK,let`s see what we can combine If I want to create the SATL and FATL indicators, I think I've to use the short periods for FATL and the long periods for SATL. Am I right? Is it ok for example to take P1=73 and D1=30 for SATL and P1=12 and D1=21 for SATL?Yes,it is ok,this is the right concept.

And in this case, I should not divide the values by 2, like for a normal stochastic or RSI indicator?No,absolutely not.You only divide values by 2 when you use them in traditional oscillators,basically,for example,if cycle is 50,half cycle is 25,if you use a williams % range oscillator,,for example,you want to fit it to 25 periods,the half cycle,so it will give you the highest percentual readings at the top of the cycle(the top of the UP half),and the lowest values at the bottom of the cycle(the bottom of the down half).When you use noise filtering devices(SATL,FATL,even a traditional sma) you want to filter OUT the noise outside the dominant cycles,so you use the full cycle(s) period(s),never the half cycles...do a visual check..use a 73 periods sma delayed by 36 periods..and a 30 periods sma delayed by 14 periods..and check if they represent the past price history accurately or not..probably they will,only issue will be that you will have a lag,since smas will end 36 and 14 points ago....that is why digital filters are very useful..their lag is minimal.

About the filters' sets, how do we combine the filters if I want to create a STLM - FTLM indicator for example?This is more advanced,stlm2&ftlm use double buffering to achieve additional smoothness,let`s stick to the basics first,once you know how to do a customized satl or a fatl,we can proceed,and then it will just be a matter of copying and pasting,onceyou understand the concept and the filter mechanics,it becomes easy.

Regards,

Daniel

Hi Daniel,as usual ,the reply is above ..Regarding STLM2 and FTLM,I will explain once you can manage SATL and FATL..anyway,I have to tell you that STLM2 and FTLM are already very robust indicators (even more than SATL),and it is very difficult to find a pair and timeframe where it is worth the effort to build a customized STLM2/FTLM,anyway,we will do it if you want..as a next step,once you can handle SATL and FATL as second nature.

Regards

Simba

[dvarrin]

[SIMBA]

[dvarrin]

SIMBA:
Hi,now the most interesting part...please read my comments above and be so kind to post some charts where you put the different digital filters on the pair and timeframe that you selected for analysis..Nothing like seeing the results on an actual chart

Hi Simba,

Here is the chart with SATL, FATL and PCCI.

I've tried with the values you gave me and it was ok for FATL and PCCI. I had no reason for taking 31 instead of 29

I've also noticed that depending on which value I'm using for D1, the curve can be offset compared to the price. Is it a bug in the DFM tool?

I'm longing to know more about the other more complex indicators we can create.

[SIMBA]

D1

dvarrin:
Hi Simba,

Here is the chart with SATL, FATL and PCCI.OK..looks like you have a way NOW to define the trend for EURUSD H4..for example PCCI above 0 plusFATL above SATL and SATL slope is positive..looks like a good h4 uptrend that you can trade with oscillators,RBCI,STLM2,etc in h1,m30 or m15

I've tried with the values you gave me and it was ok for FATL and PCCI. I had no reason for taking 31 instead of 29 OK

I've also noticed that depending on which value I'm using for D1, the curve can be offset compared to the price. Is it a bug in the DFM tool?NO..it is just the fact that you are changing the cutoff period..try to do an alternative SATL with p1=63..d1=29 and compare with your actual one,it will probably be faster but less smooth...but probably useful too..try to do it and post a pic and compare

I'm longing to know more about the other more complex indicators we can create.Yes,understandably so...and you will realize that most of them are based on SATL and FATL

Hi Daniel,

As usual,my comments are located above...try to decide which of the SATLs and FATLs suits better your trading style,and then we will proceed to STLM2 and FTLM..but first,we will need to stop at RSTL and RFTL.To understand the concepts behind RFTL & RSTL it is important that you read my explanation about centered moving averages at the SimbaConMan thread,or,even better ,alternatively,read the J.M Hurst book.

[clahn04]

for some reason i cannot get the data that i export out of metatrader (csv) to load in the spectral analyzer...i get an error....can someone walk me through that process?

thanks,

cl

edit- i figured it out nevermind

[dvarrin]

SIMBA:
Hi Daniel, As usual,my comments are located above...try to decide which of the SATLs and FATLs suits better your trading style,and then we will proceed to STLM2 and FTLM..but first,we will need to stop at RSTL and RFTL.To understand the concepts behind RFTL & RSTL it is important that you read my explanation about centered moving averages at the SimbaConMan thread,or,even better ,alternatively,read the J.M Hurst book.

Hi Simba,

I've read the whole SimbaConMan thread. I've seen the centered moving averages method. It looks really interesting. What is the relation between that method and RSTL and RFTL?

[SIMBA]

dvarrin:
Hi Simba, I've read the whole SimbaConMan thread. I've seen the centered moving averages method. It looks really interesting. What is the relation between that method and RSTL and RFTL?

If you read the Hurst book,you will realize that the difference between a cycle and itself delayed by half the cycle period,in a perfect theoretical world,will nail the tops and the bottoms of the cycle..in real life,it does a fairly good job to approach tops and bottoms,the problem being the time delay inherent in the use of centered(delayed by half) moving averages,and the fact that cycles are non stationary,unless you can find some cycle with very high Bartels.

To solve the first problem problem,we can use a SATL and a SATL delayed by half its period(RSTL)and estimate the difference(that is what the STLM2 does,BTW),so we have a "conceptual Hurst " method in real time...and now the question..check the standard SATL and the standard RSTL..Is it delayed by half the period?

[dvarrin]

SIMBA:
If you read the Hurst book,you will realize that the difference between a cycle and itself delayed by half the cycle period,in a perfect theoretical world,will nail the tops and the bottoms of the cycle..in real life,it does a fairly good job to approach tops and bottoms,the problem being the time delay inherent in the use of centered(delayed by half) moving averages,and the fact that cycles are non stationary,unless you can find some cycle with very high Bartels. To solve the first problem problem,we can use a SATL and a SATL delayed by half its period(RSTL)and estimate the difference(that is what the STLM2 does,BTW),so we have a "conceptual Hurst " method in real time...and now the question..check the standard SATL and the standard RSTL..Is it delayed by half the period?

Is it the definition of RSTL that it is delayed by half the period of SATL? I tried to put both SATL and RSTL on a chart, but the offset is not exactly half the period.