| Welcome Guest, please sign in to participate in a discussion. | Search | Active Topics | |
Registered User
Joined: 1/31/2005
Posts: 24
|
In working with the Detrended Price Oscillator in Blocks, I notice that the DPO plot is about a week behind prices shown. It appears that the DPO is offsetting data by that period? On other charts of DPO that I have seen, DPO appears currently with price, and no lag. Am I missing something here?
|
|
Worden Trainer
Joined: 10/1/2004
Posts: 18,819
|
I am not familiar enough with DPO to say if this is normal. Let me do some research and get back to you here.
- Craig
Here to Help!
|
|
Registered User
Joined: 1/31/2005
Posts: 24
|
Hi Craig - any luck with your DPO research? I understand the concept - just don't understand the lag in the plot of DPO.
|
|
Registered User
Joined: 1/28/2005
Posts: 6,049
|
As I understand it. DPO is equivalent to
the close minus its moving average.
(half the period plus 1 bars ago)
In telechart a DPO of 20 would be:
C-AVGC20.11
(the close minus its 20 period moving average
11 bars ago: (20*.5)+1)
Thanks
diceman
|
|
Registered User
Joined: 1/1/2005
Posts: 2,645
|
phillrun,
I am guessing, but I believe that you are refering to the DPO not being plotted near the current date rather than any lag with respect to the Price plot. This is because the DPO as plotted in Blocks requires future data.
The DPOs of period P as provided by the references are:
1) MarketScreen:
C - AVGCP.(P/2+1)
2) Paritech:
C - AVGCP.(-(P-1)/2)
3) Paritech & Wikipedia:
C - AVGCP.(-(P/2+1))
4) Blocks:
C - AVGCP.(-(P/2-2))
Notes:
1) is the only one that can be plotted current as it does not require future
data. It does not attenuate cycles with period greater than P as claimed.
2) seems to be what all strive for but none achieve. Paritech only alludes to it. It is the only one that does not introduce phase distortion.
3) is introduced for some reason unknown to me. It appears to have a little better ripple suppression than 2) at the expense of phase distortion.
4) is how I interpret the code, with which I am not familiar. It appears to be a coding error. Does the plot stop P/2-2 bars from the current bar when P is even and P >= 4?
Hopefully, someone will verify 4).
Thanks,
Jim Murphy
|
|
Registered User
Joined: 1/31/2005
Posts: 24
|
Jim - Thanks for the formulae. You are correct - the DPO in blocks stops p/2-2 bars from the current day. Is this a coding error, or is it just a fact that the signals on DPO will always be a few days late? Ok either way, but need to know.
Also, is the formula for DPO in Telechart as given by Craig have a lag also or is it plotted to current date? Doesn't look like it requires future data.
|
|
Registered User
Joined: 1/1/2005
Posts: 2,645
|
QUOTE (bustermu) The DPOs of period P as provided by the references are:
1) MarketScreen:
C - AVGCP.(P/2+1)
2) Paritech:
C - AVGCP.(-(P-1)/2)
3) Paritech & Wikipedia:
C - AVGCP.(-(P/2+1))
4) Blocks:
C - AVGCP.(-(P/2-2))
Notes:
1) is the only one that can be plotted current as it does not require future
data. It does not attenuate cycles with period greater than P as claimed.
2) seems to be what all strive for but none achieve. Paritech only alludes to it. It is the only one that does not introduce phase distortion.
3) is introduced for some reason unknown to me. It appears to have a little better ripple suppression than 2) at the expense of phase distortion.
4) appears to be a coding error.
I am going to modify my statement:
4) appears to be a coding error.
and replace it by:
If 4) is a coding error, then it may be a good one.
This is because the filter of 4) may be preferable to that of 3).* The real question to be answered is whether either is preferable to the filter of 2). Until then, I would advise using the filter of 2).
*The reason for this is that the frequency response of one is the complex conjugate of that of the other and the filter of 4) can be plotted three bars more recently than the filter of 3).
Thanks,
Jim Murphy
|
|
|
Guest-1 |