[walterw]

About the Fisher Oscillator I did some testings so far and the lime / magenta dots have re-painting issues, making them non reliable... as to the oscillator himself, he doesnt have any re-painting issues and I must confess he is a nice decent oscillator... will keep testing him... I understand he is based on Ehlers work about the fisher transform algorithm... thanks Agustin for this inputs... cheers Walter.

[walterw]

uups... re-painting issues on the fisher oscillator as well... lol

[cattus]

Originally Posted by walterw »

About the Fisher Oscillator I did some testings so far and the lime / magenta dots have re-painting issues, making them non reliable... as to the oscillator himself, he doesnt have any re-painting issues and I must confess he is a nice decent oscillator... will keep testing him... I understand he is based on Ehlers work about the fisher transform algorithm... thanks Agustin for this inputs... cheers Walter.

Hi Walter.

Thanks to take this oscillator to test it.

I added the signals Lime/Magenta. I am not a coder, that was the reason by that I asked for some coder here to review the program in order to try to solve that kind of problem. May be PYenner or Blu-Ray can help us with this problem.

My original intention is to try to use it indicator do not for timing, I want to use just to confirme the inflections. I have observed that sometimes the Chimp indicator can show us an inflection but Fisher doesn't show us the same inflection.

Other advantage for which I want to use this indicator is that sometimes Fisher shows the inflection before it occurs, then, we could prepare in order to confirm with the Chimp and the cross of NonLag over the ADXVMA (Lime line).

The issue of timing for me is as you said it. But obviously, it can evolve...

Fisher comes from Ehlers works, as you mentioned.

Thanks.

cattus

[walterw]

Originally Posted by cattus »

Other advantage for which I want to use this indicator is that sometimes Fisher shows the inflection before it occurs,

It plots historically that way, but on real time he is no better than ccit3 wich I find good if you wanted to time on 5 minutes...

So far for cycle detection chimp is ok, if you wanted to time the inflections on 5 min ccit3 its about one of the best oscillator for that...

I post later on that issue... cheers Walter.

[cattus]

Originally Posted by walterw »

It plots historically that way, but on real time he is no better than ccit3 wich I find good if you wanted to time on 5 minutes...

So far for cycle detection chimp is ok, if you wanted to time the inflections on 5 min ccit3 its about one of the best oscillator for that...

I post later on that issue... cheers Walter.

Hi Walter.

I downloaded the CCIT3 in order to test it, it seems very good.

The main problem that I see in Fisher indicator is the re-paint, and I believe it is an implicit problem of the original formula, it is not a problem of the signals (Lime/Magenta) that I added.

All your comments about it indicator are very important to me. I guess the comments of coders will be good for us.

Thanks again.

cattus

[cattus]

Hi Walter.

I would like to check with you the inputs of the lines on 5 min chart. The name of the indicators over the price chart are:
ADXVMA6_16bar
ADXVMA6_12bar
TLOMI NonLagMA_v4

When I use the template you attached before, called "Chimp 5 min chart" and after that I check the inputs of the ADX's lines, I find these inputs:
ADXVMA6_16bar (the input for ADX_Bars is 40)
ADXVMA6_12bar (the input for ADX_Bars is 4)
TLOMI NonLagMA_v4 (inputs are 0,9,0,0,0,1,0.0)

I don't see any problem at TLOMI NonLagMA_v4, but the name of the others two tell me that the inputs must be 16 and 12.

What are the correct inputs?:
ADXVMA6_16bar: 16? or 40?
ADXVMA6_12bar: 12? or 4?

Note: I asume that you know that with a single indicator called ADXVMA6 we have the source to put any ADXVMA6 with different inputs, but may be you add at each indicator the annex _16, _12, _6, etc. for not confuse us.

Thanks for your help.

AgustÃÂ*n

[walterw]

so far it is 40 and 4... names are mixted... sorry about that... little messy research on here...


Cattus : since your inputs about timeframes, I had been looking arround a 60 min against a 5 min thats 1/12th relation between timeframes... it could call for more robust scalps at 5 min (still good lifestyle issues under control) with a very solid trend definition at 60 min in this case using a rainbow for good momentum determination...

Must confess we have so many combinations with vma`s and I am working almost 14 hours per day testing combinations... we will get to the best one... I just dont want to loose interaction with you guys, thats why I started this last thread, dont want to loose momentum on the interaction process as I
am learning together with all of you... thanks so much for inputs... even when they turn out "maybe" futile... its 1 lamp test less

Edison tested 3000 bulbs until success... even thought he was deaf..
so I find refreshing all your interactions...

About oscillators Cattus, ccit3 its the first on my list after extensive research... chimp looks good for cycle recognition, not timing...

we keep this process ongoing...

One thing somehow its finished for me.. its the "setuping" issues with the centrifugal and centripetal setups... thats a finished topic for me... what context of indicators and time frames, I believe there is more to be done on optimization...

one nice frase from Edison : "Genius is one percent inspiration, ninety-nine percent perspiration." - Thomas Alva Edison

here is a nice link about him http://en.wikipedia.org/wiki/Thomas_Edison very nice inspiration from this guy...


cheers Walter.

[unicorn]

Originally Posted by walterw »

uups... re-painting issues on the fisher oscillator as well... lol

Hello Walter and Cattus;

The implementation of the fisher oscillator that you are using is incorrect.
This implementation goes back to the "step rsi" and "step stochastics" indicators, that start the computation loops at the last bar and propagate the calculations to the past. This is a trick that makes the step indicators look good but of course has the result of re-computation and thus "re-painting".

The iFisher stochastic uses the inverse Fisher transform
inverse_fisher_value = 0.5 * log ( (1 + osc_Value) / (1 - osc_Value) ) ;

on the stochastic oscillator after it has been rescaled to have values in the range (-1, +1)
osc_value = 2 * stoch(fast_periods, slow_periods) -1;

It is a stochastic that has been stretched a little bit. Similar to iFisher_rsi as explained by John Ehlers at his site.

If you like try the attached implementation (developed by member "feb2006" at the site where you probably got your mq4). Let us know your evaluation.