| leegyoung |
|
Gold User, Member, TeleChart
|
| Registered User |
|
|
|
|
| Unsure |
|
| Monday, October 18, 2004 |
| Sunday, January 12, 2020 11:06:52 PM |
90
[0.03% of all post / 0.01 posts per day] |
|
|
Have attempted to use the Moving Linear Regression Line as a scan condition, but am unable to do so; i.e. when I scanned U.S. Stocks, nothing came up although I found several by scanning manually; used the same setup for a Hull Moving average; found many stocks?????
Am Using Ver. 17.0.6283.28984
|
I was working with synthesizing George Appel's MACD back in October/December 2004 and want to take up this work again.
On December 14, 2004 one of your trainers stated the following:
"To get a 60-day XMA of the exponential MACD, you need to be a little careful, since there is a weird bug in the current version that calls for nested XMA's to keep the innermost term the highest one....otherwise the accuracy of the resuslt is significantly affected in many cases. The formula you provided (with a X in front of each AVG) *should* work properly once the bug is fixed."
Does this all make sense and does this "bug" still exist?
 Lee
|
You are correct; I was mixing simple and exponential.
Thanks
|
|
That form is what I started with. It is only when I put a bias of 0.3-0.4 that I begin to significantly reduce the "errors" in my Percent True.
|
I am sitting here looking at a ten day MA and a fifty day MA, both exponential. I am comparing them with a Percent True indicator using the following boolean equation: (AVGC10)>(AVGC50) AND AVGC50<(AVGC10); the two do not track. That is to say, the Percent True indicator, which is designed to be "high" when the MA10 is greater than the MA50, will sometimes appear "high" when the MA50 comes down close to and just touching but not going below the MA10.
Is this just a graphical resolution problem? Got any ideas?
Lee
|
When using the "Regressions to Indicator and Price" option on Moneystream, what are the period intervals that are used; or are they "self-fitting?"
Lee
|
Can you suggest a PCF in which to determine the trend (i.e. slope) of Price ?
Lee
|
Thanks Jim and Craig,
Craig, I'll view the video; have not done so previously in that I assumed it would be based upon the STOC established within TC2000. That would not work because I need access to %K and %D curves in order to compare them to each other within the Percent True Indicator.
At the minimum, I can write the %K and %D equations using "brute force."
However, before I do so I'll go through your comments Jim.
Thanks to both of you!
Lee
|
Hi Craig,
Thanks for the fast response!
Since that approach isn't possible, let me change the question as follows; I'm attempting to define individual
lines, %K and %D, for both the Fast and Slow Stochastic. I wish to use them as individual crossing indicators in conjunction with the Percent True indicator.
The functions look like this;
%K=100*[(C-MINLN)/(MAXHN-MINLN)]
Where N can be a large number of periods.
%D=100*[M Days summation of (C-MINLN)/M Days Summation of (MAXHN-MINLN)]
Where M can also be a large number of days, separate from N.
The majority of either algorithim is simple; I just don't know how to sum a series of quotients as above.
I realize that TC200 has a Stochastics function, but I would like to use the above functions in the Percent True Indicator. I.e. "If %D >=%K then..."
Can this be accomplished with TC2000?
Lee
|
In order to define a particular form of Stochastic, literally the individual lines for %K and %D, I wish to,
1) calculate, for use as an indicator, the five period Exponential Moving Average of a user defined PCF; is the below the correct syntax?
xavg(PCF),5; and,
2) the five period Exponential Moving Average of the function above;
is the below the correct format?
xavg(xavg(PCF),5),5
Lee
|
|