Download software Tutorial videos
Subscription & data-feed pricing Class schedule


New account application Trading resources
Margin rates Stock & option commissions

Attention: Discussion forums are read-only for extended maintenance until further notice.
Welcome Guest, please sign in to participate in a discussion. Search | Active Topics |

Converging DPO Rate this Topic:
Previous Topic · Next Topic Watch this topic · Print this topic ·
farr_west
Posted : Tuesday, October 17, 2006 8:05:10 PM
Registered User
Joined: 2/2/2005
Posts: 9
Last year Doug and Steve helped me learn how to create a custom indicator for Detrended Price Oscillator (DPO). Now I would like to know how to create a PCF that I can use in an easy scan. I would like to take the formula C - AVGC15.8 and 2 other time periods of the same formula and determine if they are converging or close to converging.

I have this set-up on a custom indicator, but I would avoid looking through all the charts to find the ones that are converging or close to converging.

Can you help?

Thanks,

Farr West
diceman
Posted : Tuesday, October 17, 2006 9:01:55 PM
Registered User
Joined: 1/28/2005
Posts: 6,049
farr west

If I am understanding you:

Try:

ABS( DPO1-DPO2 )+ABS( DPO2-DPO3 ) and so on. Large values

will result when DPO's are far apart. Small values when they are close.

Here is an example with moving averages:

How to find stocks where the 10,20,50,100 and 200 MA's are close together?

Thanks
diceman
farr_west
Posted : Tuesday, October 17, 2006 11:18:00 PM
Registered User
Joined: 2/2/2005
Posts: 9
Diceman,

I tried that and it unfortunately it does not do what I am trying to do. I had one stock with a small number that was not convergent and another with a number 3X as big that is convergent. I am looking for a boolean "TRUE" result that I can use in an easy scan. On my chart, I have 3 DPOs (in one window) and often when they are postively converging/convergent, I have a very good buying opportunity. Any other ideas?

Farr West
diceman
Posted : Wednesday, October 18, 2006 12:56:29 AM
Registered User
Joined: 1/28/2005
Posts: 6,049
Realize they will each be on their own scales. They my look like they
are in agreement but may not be. (that could be why your ABS scan seemed not to work.
--------------------------------------------------------------------------
I would try taking the shortest period DPO and the longest period DPO.
Apply a liner regression line to each and do a sort based on their relationship.
(a difference sort)
--------------------------------------------------------------------------
You may also want to try the short minus long DPO's as an oscillator.

Obviously when they are crossing the zero line. they are close together.

You would also be able to see the expansion/contraction of their relationship.

Thanks
diceman

diceman
Posted : Wednesday, October 18, 2006 1:28:44 AM
Registered User
Joined: 1/28/2005
Posts: 6,049
This is another idea to try.
I am assuming the DPO is:

C-AVGC20.10 (you can adjust it to your real values)
------------------------------------------------------------------------------------
SCAN:

(C-AVGC20.11>C4-AVGC20.15)AND(C4-AVGC20.15<C14-AVGC20.25)

This is basically asking if DPO today is > 4 bars ago and is 4 bars ago
less than 10 bar before that.

(numbers I made up for a DPO turning point you can try to make the values better)

Thanks
diceman
farr_west
Posted : Thursday, October 19, 2006 7:56:34 PM
Registered User
Joined: 2/2/2005
Posts: 9
Diceman,

You are talking about changing my formula.

I want the formula to stay the same, except for the "15.8" portion. It could be 24.13, 33.17, 45.23, etc. Anything that user wants to plug in. I could not figure it out and was hoping a trainer could.

In essence, I need to know if there is a way to program something like this:

C-AVGC15.8 is equal to or within 5% of C-AVGC24.13 which is equal to or within 5% of C-AVGC45.23

On the chart this will look 3 line indicators all coming to a single point, or close to.

Any ideas? Steve? Doug?
diceman
Posted : Thursday, October 19, 2006 8:24:27 PM
Registered User
Joined: 1/28/2005
Posts: 6,049
You can edit my formula to your DPO values.
-------------------------------------------------------------------------------------
Quote:"In essence, I need to know if there is a way to program something like this:

C-AVGC15.8 is equal to or within 5% of C-AVGC24.13 which is equal to or within 5% of C-AVGC45.23 "

The problem with 5% is the values are going above and below the zero line.
(negative and positive numbers)

If we compare ABS values we are back to the first post.

----------------------------------------------------------------------------------------------------------
Have you tied my scan as a % true indicator while observing a price
chart. The idea is to find a turning points from a bottom?

I think this "style" has merit.

Look at the intent don't get caught up in the values

Just because I used C-AVGC20.10 does not mean the concept
is invalid.

Thanks
diceman

diceman
Posted : Thursday, October 19, 2006 9:08:06 PM
Registered User
Joined: 1/28/2005
Posts: 6,049
I would re-read what I said above.

You can experiment with this:

((ABS(C-AVGC15.8)+ABS(C-AVGC24.13)+ABS(C-AVGC45.23))/AVGC45)*100<5

You can experiment with the 5 value to make
it more or less sensitive.

Thanks
diceman
bustermu
Posted : Friday, October 20, 2006 2:15:08 AM
Registered User
Joined: 1/1/2005
Posts: 2,645
Farr West,

My interpretation of your request is that you would like to know when the three DPOs:

C-AVGC15.8, C-AVGC24.13, C-AVGC45.23

are close together.

A PCF for their maximum distance apart as a percent of the average of the three moving averages is:

100*(3/2)*(ABS(AVGC15.8-AVGC24.13)+ABS(AVGC24.13-AVGC45.23)+ABS(AVGC45.23-AVGC15.8))
/(AVGC15.8+AVGC24.13+AVGC45.23)

You will have to determine for yourself the limit to put on the PCF in an EasyScan. The value appears to me to be around 1. You would probably be better off just sorting a WatchList by the PCF.

As mentioned by diceman, you cannot determine when the DPOs are close together by observing their plots as Custom Indicators because they will be on different scales. If you plot the three delayed moving averages, AVGC15.8, AVGC24.13, AVGC45.23, as Custom Indicators with "Plot using price scale" checked, then their distance apart will be the same as the DPOs' distance apart.

Warning: The above form of the PCF does not extend directly to four or more DPOs. The form has to change.

Thanks,
Jim Murphy
Craig_S
Posted : Friday, October 20, 2006 3:14:54 PM


Worden Trainer

Joined: 10/1/2004
Posts: 18,819
farr_west,

Do you have these three custom indicators plotted in the same window? Note that they are on different scales. Any converging or crossing you see for these custom indicators is an illusion... each are on an independent scale.

The forumula bustermu gave above will find them converging BUT you may not get confirmation of this on the chart.

- Craig
Here to Help!
farr_west
Posted : Friday, October 20, 2006 6:53:55 PM
Registered User
Joined: 2/2/2005
Posts: 9
Craig, Jim, & Diceman,

I appreciate your help. I just checked and found that I have the center line box checked. I tried checking the 'plot using price scale' block and it completely disappeared off the screen. I tried multiplying by a factor, but to no avail.

So what you are telling me is that I am looking at am optical illusion and therefore there is no way to write a PCF that shows me this 'visual' convergence I see on the screen. That I will just have to manually look for charts that have these three indicators, in one window, visually converging, which happens when the stock is moving up...bummer

I don't understand this 3 different scales thing. I see a single scale onscreen (right side), that is typically between -1.00 and 1.00. Is that a percentage? Why is not it all relative? or is it?

Thanks for help,

Paul "Farr West"
Craig_S
Posted : Friday, October 20, 2006 7:45:28 PM


Worden Trainer

Joined: 10/1/2004
Posts: 18,819
When you add a custom indictator in TeleChart it brings its own scale. It scales from the highest to the lowest values for its own plot.

Imagine each one being drawn on its own piece of clear plastic. Each independent of the other, each range of values drawn from the top of their plastic piece to the bottom. When you put all three in the same window, all you are doing is placing these plastic sheets on each other. This does not change the values at the top and bottom of every sheet; each indicator has its own scale and range of values. You can slide these sheets up and down relative to each other and it will change any crossovers or converging you see.

Any visual convergence is not a mathmatical one... changing the zoom of you chart will show you how the indicators can change relative to each other.

What you are likely trying to uncover is a mathmatical convergence (which makes some technical sense). Try bustermu's formula:

100*(3/2)*(ABS(AVGC15.8-AVGC24.13)+ABS(AVGC24.13-AVGC45.23)+ABS(AVGC45.23-AVGC15.8))
/(AVGC15.8+AVGC24.13+AVGC45.23)

Sort a few WatchLists by this PCF in ascending and descending order. See if you like the results.

Note the scale you see on the right is color coded to the Custom Indictator. Click the grey box at the bottom of the scale to switch to the scale ranges of the other indicators in the window.

- Craig
Here to Help!
farr_west
Posted : Wednesday, October 25, 2006 9:30:11 PM
Registered User
Joined: 2/2/2005
Posts: 9
"I see" said the blind man. I tried a couple of things.

1. I looked at what was some of the other criteria that accompanied the "visual convergence" of my 3 DPO indicators and have setup (still tweaking it) an Easy Scan that attempts to find opportunities this way. Would love to hear others thoughts.

2. Since I am looking for some sort of a "go-nogo" gauge. I came up with the following PCF:

ABS(C - XAVGC15.8) > ABS(C - XAVGC30.16 * 0.95) OR ABS(C - XAVGC15.8) < ABS(C - XAVGC30.16 * 1.05) AND ABS(C - XAVGC15.8) > ABS(C - XAVGC100.51 * 0.95) OR ABS(C - XAVGC15.8) < ABS(C - XAVGC100.51 * 1.05)

I am still playing with the parameters, but I am encouraged because I only look at stocks that return a True result. Then I can take the PCF Bustermu provided and rank them. I'd like to hear any feedback on this approach.

Thanks,
Farr West
Users browsing this topic
Guest-1

Forum Jump
You cannot post new topics in this forum.
You cannot reply to topics in this forum.
You cannot delete your posts in this forum.
You cannot edit your posts in this forum.
You cannot create polls in this forum.
You cannot vote in polls in this forum.