raven4ns
23 posts
msg #33666
- Ignore raven4ns |
10/24/2004 4:09:57 PM
Hello,
I've been using the T3 MA to help time entry and exit. It's not too bad. Apparently, Jurik has a MA that is smoother then the T3. However, I saw a moving average using the MEF (Ehler's filter) and it was almost as good but didn't cost $200 usf to get. Can anyone write a filter using the MEF for a 3 day and a 5 day MA? I would like to compare it with the T3.
For those of you who are using the T3 you might want to increase the variable from .70 to .84. I saw some work done on this and when I checked it out this seemed to be about the best IMHO. Thank you.
Tim
|
yepher
359 posts
msg #33671
- Ignore yepher |
10/24/2004 11:33:43 PM
raven4ns,
Sorry, I tried to write the filters for MEF and MTI filters from the Optimal Detrending article by John Ehlers.
Unfortunetly I could not get the filter to work within the constraints of the Stockfetcer "set" command limit which I believe only allows nested variables up to 4 levels deep.
I tried several approaches to try and reduce the nesting level but none worked for me. Maybe someone who is really creative will be able to overcome this.
Anyone see how to get this to work in less then four layers of set{}?
High + Low
Price = -----------
2
Smooth = (0.13784 * (2*Price[0] - Price[-1])) +
(0.0007 * (2*Price[-1] - Price[-2])) +
(0.13785 * (2*Pirce[-2] - Price[-3])) +
(1.2103 * (Smooth[-1] - (0.4867 *Smooth[-2]))
If this one can be done in less than 4 nested sets then the rest of the indicators can be implemented also.
Maybe the nice Stockfetcher Staff will add this indicator for all to use :)
SF,
If you are interested in implementing this for us you can find the indicator in the magazine Technical Analysis of Stocks & Commidities, Volume 18 by John Ehlers 2000.
|
yepher
359 posts
msg #33672
- Ignore yepher |
10/24/2004 11:34:55 PM
In the previouse post the spacing got messed up a little. Here is how "Price" should have read:
Price = (High + Low) / 2
|
cegis
235 posts
msg #33675
- Ignore cegis |
10/25/2004 6:19:39 PM
yepher,
The limit to levels of nesting of set{} is not a hard-and-fast 4 levels. It depends on the functions that the set{}s use as well ("total computational expense" is the limit). Please post what you have, and I'll see if I can help with getting it within the SF limit.
C
|
cegis
235 posts
msg #33676
- Ignore cegis |
10/25/2004 7:03:07 PM
On second look, the formula is based on the prior day's formula value. I don't think you'll be able to do that with set{}s... (Assuming "[-1]" means 1 day ago.)
I'll still take a look, if you want to post what you have, tho...
C
|
yepher
359 posts
msg #33681
- Ignore yepher |
10/26/2004 12:43:03 AM
Thanks for the offer Cegis but I thew away what I had because I could not get it to work.
I think this particular filter would be better if StockFetcher would add it as a standard indicator.
|
guru_trader
485 posts
msg #51710
- Ignore guru_trader |
5/20/2007 6:19:20 PM
Is there still a problem in writing a filter for this type of indicator?
|