Registered User
Joined: 5/9/2011
Posts: 80
|
Hello,
I am writing a scan that requires the values of the %Price Oscillator both in absolute values and comparative in Boolean expression. Plus, I need to write the equation for the Linear Regression of this indicator too for the periods of 10, 7, 5, 3, and 2...
Lastly can you supply the syntax for finding the values moving back a day at a time.
Thank you,
Jim
|
|
Worden Trainer
Joined: 10/7/2004
Posts: 65,138
|
The formula for a PPO is fairly straightforward given the settings. For example, a PPO 12,26 could be written as follows.
100 * (XAVGC12 / XAVGC26 - 1)
But we do need to know the settings to write the formula.
Note that while formula for the linear regression of the raw PPO can probably be written, it is probably not possible to do the same for a moving average of PPO or for a PPO Histogram.
-Bruce
Personal Criteria Formulas
TC2000 Support Articles
|
|
Registered User
Joined: 5/9/2011
Posts: 80
|
Hi Bruce,
Settings for the PPO (5,35,9)...
My goal is to calculate the LR for the Histogram. And, to find the days that the PPO crosses the axis in either direction on any given day?
Thanks, Jim
|
|
Registered User
Joined: 5/9/2011
Posts: 80
|
Bruce will you have feedback on my question above? Thanks Jim
|
|
Worden Trainer
Joined: 10/7/2004
Posts: 65,138
|
The PPO 3,35 is going to look familiar.
100 * (XAVGC5 / XAVGC35 - 1)
But the PPO Histogram 5,35,9 is quite long.
100 * (XAVGC5 / XAVGC35 - .20019827 * (XAVGC5 / XAVGC35 + .8 * (XAVGC5.1 / XAVGC35.1 + .8 * (XAVGC5.2 / XAVGC35.2 + .8 * (XAVGC5.3 / XAVGC35.3 + .8 * (XAVGC5.4 / XAVGC35.4 + .8 * (XAVGC5.5 / XAVGC35.5 + .8 * (XAVGC5.6 / XAVGC35.6 + .8 * (XAVGC5.7 / XAVGC35.7 + .8 * (XAVGC5.8 / XAVGC35.8 + .8 * (XAVGC5.9 / XAVGC35.9 + .8 * (XAVGC5.10 / XAVGC35.10 + .8 * (XAVGC5.11 / XAVGC35.11 + .8 * (XAVGC5.12 / XAVGC35.12 + .8 * (XAVGC5.13 / XAVGC35.13 + .8 * (XAVGC5.14 / XAVGC35.14 + .8 * (XAVGC5.15 / XAVGC35.15 + .8 * (XAVGC5.16 / XAVGC35.16 + .8 * (XAVGC5.17 / XAVGC35.17 + .8 * (XAVGC5.18 / XAVGC35.18 + .8 * (XAVGC5.19 / XAVGC35.19 + .8 * (XAVGC5.20 / XAVGC35.20 + .8 * (XAVGC5.21 / XAVGC35.21 + .8 * (XAVGC5.22 / XAVGC35.22 + .8 * (XAVGC5.23 / XAVGC35.23 + .8 * (XAVGC5.24 / XAVGC35.24 + .8 * (XAVGC5.25 / XAVGC35.25 + .8 * (XAVGC5.26 / XAVGC35.26 + .8 * (XAVGC5.27 / XAVGC35.27 + .8 * (XAVGC5.28 / XAVGC35.28 + .8 * (XAVGC5.29 / XAVGC35.29 + .8 * (XAVGC5.30 / XAVGC35.30))))))))))))))))))))))))))))))))
Cascades of Moving Averages
Which is why I indicated in my original response that formulas for the linear regression of the PPO Histogram would not be possible. At least I can't post them here as they are too long and slow to be practical or post in the forums.
Using Linear Regression vs Classical Peaks/Valleys for Divergence Analysis
A Condition Formula for the PPO Histogram crossing through zero in either direction is about twice as long as the Indicator Formula for the PPO Histogram.
SGN(XAVGC5 / XAVGC35 - .20019827 * (XAVGC5 / XAVGC35 + .8 * (XAVGC5.1 / XAVGC35.1 + .8 * (XAVGC5.2 / XAVGC35.2 + .8 * (XAVGC5.3 / XAVGC35.3 + .8 * (XAVGC5.4 / XAVGC35.4 + .8 * (XAVGC5.5 / XAVGC35.5 + .8 * (XAVGC5.6 / XAVGC35.6 + .8 * (XAVGC5.7 / XAVGC35.7 + .8 * (XAVGC5.8 / XAVGC35.8 + .8 * (XAVGC5.9 / XAVGC35.9 + .8 * (XAVGC5.10 / XAVGC35.10 + .8 * (XAVGC5.11 / XAVGC35.11 + .8 * (XAVGC5.12 / XAVGC35.12 + .8 * (XAVGC5.13 / XAVGC35.13 + .8 * (XAVGC5.14 / XAVGC35.14 + .8 * (XAVGC5.15 / XAVGC35.15 + .8 * (XAVGC5.16 / XAVGC35.16 + .8 * (XAVGC5.17 / XAVGC35.17 + .8 * (XAVGC5.18 / XAVGC35.18 + .8 * (XAVGC5.19 / XAVGC35.19 + .8 * (XAVGC5.20 / XAVGC35.20 + .8 * (XAVGC5.21 / XAVGC35.21 + .8 * (XAVGC5.22 / XAVGC35.22 + .8 * (XAVGC5.23 / XAVGC35.23 + .8 * (XAVGC5.24 / XAVGC35.24 + .8 * (XAVGC5.25 / XAVGC35.25 + .8 * (XAVGC5.26 / XAVGC35.26 + .8 * (XAVGC5.27 / XAVGC35.27 + .8 * (XAVGC5.28 / XAVGC35.28 + .8 * (XAVGC5.29 / XAVGC35.29 + .8 * (XAVGC5.30 / XAVGC35.30)))))))))))))))))))))))))))))))) <> SGN(XAVGC5.1 / XAVGC35.1 - .20019827 * (AVGC5.1 / AVGC35.1 + .8 * (AVGC5.2 / AVGC35.2 + .8 * (AVGC5.3 / AVGC35.3 + .8 * (AVGC5.4 / AVGC35.4 + .8 * (AVGC5.5 / AVGC35.5 + .8 * (AVGC5.6 / AVGC35.6 + .8 * (AVGC5.7 / AVGC35.7 + .8 * (AVGC5.8 / AVGC35.8 + .8 * (AVGC5.9 / AVGC35.9 + .8 * (AVGC5.10 / AVGC35.10 + .8 * (AVGC5.11 / AVGC35.11 + .8 * (AVGC5.12 / AVGC35.12 + .8 * (AVGC5.13 / AVGC35.13 + .8 * (AVGC5.14 / AVGC35.14 + .8 * (AVGC5.15 / AVGC35.15 + .8 * (AVGC5.16 / AVGC35.16 + .8 * (AVGC5.17 / AVGC35.17 + .8 * (AVGC5.18 / AVGC35.18 + .8 * (AVGC5.19 / AVGC35.19 + .8 * (AVGC5.20 / AVGC35.20 + .8 * (AVGC5.21 / AVGC35.21 + .8 * (AVGC5.22 / AVGC35.22 + .8 * (AVGC5.23 / AVGC35.23 + .8 * (AVGC5.24 / AVGC35.24 + .8 * (AVGC5.25 / AVGC35.25 + .8 * (AVGC5.26 / AVGC35.26 + .8 * (AVGC5.27 / AVGC35.27 + .8 * (AVGC5.28 / AVGC35.28 + .8 * (AVGC5.29 / AVGC35.29 + .8 * (AVGC5.30 / AVGC35.30 + .8 * (AVGC5.31 / AVGC35.31))))))))))))))))))))))))))))))))
I know there are problems with creating conditions based on the PPO Histogram by clicking on the chart antd selecting Create Scan Condition. You can create such conditions by using the second formula in this post in a Custom PCF Indicator as a stand in for the PPO Histogram however.
Create Conditions from Your Chart (5:25)
-Bruce
Personal Criteria Formulas
TC2000 Support Articles
|
|
Registered User
Joined: 5/9/2011
Posts: 80
|
Bruce,
Percentage Price Oscillator (PPO): {(12-day EMA - 26-day EMA)/26-day EMA} x 100
Signal Line: 9-day EMA of PPO
PPO Histogram: PPO - Signal Line
In the example above for (12,26,9) like a MACD except the resultant is divided by the long term EMA, is this the generated value that is plotted onto the histogram?
And, I understand now why the Linear Regression formula is so long.
On the Create Scan Function in TC2000, the conditions used within the available searches are capable as a function of past data in a database that performs a search on past data? And, also performing calculations on that data to illustrate in a comparable format? This is good, but it does not allow us to describe the geometry we are looking for.
Jim
|
|
Registered User
Joined: 5/9/2011
Posts: 80
|
And, is moving up and moving down smart a slope calculation? Or, is there more logic built into that condition? Thanks, Jim
|
|
Worden Trainer
Joined: 10/7/2004
Posts: 65,138
|
Yes, the formulas are like the MACD except expressed as percentages of the long moving average.
Conditions created by clicking on an indicator and selecting Create Scan Condition are for the current bar. You can add 1 period moving average to the indicator, set the Offset of the moving average to the number of bars ago of interest and selecting Create Scan Condition to check for the condition being true offset bars ago.
That is about the best you are going to be able to do. TC2000 doesn't have any built in backtesting features. You can search the history like a database to find out all of the instances where it was true.
About the closest you can get is a sort of visual backtesting using Custom PCF % True Indicators. But you have to actually be able to write the desired Condition Formula to do so. This isn't going to happen with linear regressions of the PPO Histogram because the formulas are just too long and slow.
-Bruce
Personal Criteria Formulas
TC2000 Support Articles
|
|
Registered User
Joined: 5/9/2011
Posts: 80
|
Conditions created by clicking on an indicator and selecting Create Scan Condition are for the current bar. You can add 1 period moving average to the indicator, set the Offset of the moving average to the number of bars ago of interest and selecting Create Scan Condition to check for the condition being true offset bars ag
Can you supply and example of this methodology using the Create Scan Feature?
Thank you, Jim
|
|
Worden Trainer
Joined: 10/7/2004
Posts: 65,138
|
Start by adding a Custom PCF Indicator to the chart with the following Formula.
100 * (XAVGC5 / XAVGC35 - .20019827 * (XAVGC5 / XAVGC35 + .8 * (XAVGC5.1 / XAVGC35.1 + .8 * (XAVGC5.2 / XAVGC35.2 + .8 * (XAVGC5.3 / XAVGC35.3 + .8 * (XAVGC5.4 / XAVGC35.4 + .8 * (XAVGC5.5 / XAVGC35.5 + .8 * (XAVGC5.6 / XAVGC35.6 + .8 * (XAVGC5.7 / XAVGC35.7 + .8 * (XAVGC5.8 / XAVGC35.8 + .8 * (XAVGC5.9 / XAVGC35.9 + .8 * (XAVGC5.10 / XAVGC35.10 + .8 * (XAVGC5.11 / XAVGC35.11 + .8 * (XAVGC5.12 / XAVGC35.12 + .8 * (XAVGC5.13 / XAVGC35.13 + .8 * (XAVGC5.14 / XAVGC35.14 + .8 * (XAVGC5.15 / XAVGC35.15 + .8 * (XAVGC5.16 / XAVGC35.16 + .8 * (XAVGC5.17 / XAVGC35.17 + .8 * (XAVGC5.18 / XAVGC35.18 + .8 * (XAVGC5.19 / XAVGC35.19 + .8 * (XAVGC5.20 / XAVGC35.20 + .8 * (XAVGC5.21 / XAVGC35.21 + .8 * (XAVGC5.22 / XAVGC35.22 + .8 * (XAVGC5.23 / XAVGC35.23 + .8 * (XAVGC5.24 / XAVGC35.24 + .8 * (XAVGC5.25 / XAVGC35.25 + .8 * (XAVGC5.26 / XAVGC35.26 + .8 * (XAVGC5.27 / XAVGC35.27 + .8 * (XAVGC5.28 / XAVGC35.28 + .8 * (XAVGC5.29 / XAVGC35.29 + .8 * (XAVGC5.30 / XAVGC35.30))))))))))))))))))))))))))))))))
Now add a 1-period moving average to the Custom PCF Indicator with the Offset set to 1.
Note that the indicator repeats the values plotted by the Custom PCF Indicator, it just does it 1-bar later. If you change the offset to 5, it would repeat the value from 5 bars before instead.
Let's say you wanted to find out if the formula crossed up through zero during the previous bar. Click on the offset moving average and select Create Scan Condition. Choose Crossing Up Value with Crossing Up Through set to 0.
The condition created will be for Custom PCF Indicator crossing up through 0, but during the previous bar instead of the current bar.
I referenced a video about the basic idea of clicking on an indicator and selecting Create Scan Condition in my Tuesday, March 10, 2015 3:22:50 PM ET post. The only difference is that we are adding a 1-period offset moving average and clicking on the moving average to create the condition instead of clicking on the condition itself.
-Bruce
Personal Criteria Formulas
TC2000 Support Articles
|
|
Registered User
Joined: 5/9/2011
Posts: 80
|
Thank you Bruce ... work to do this evening. Simply finding the geometrical patterns you need although many tools available, not always possible. Jim
|
|
Worden Trainer
Joined: 10/7/2004
Posts: 65,138
|
You are right. It is not always possible to identify particular patterns even if you have formulas to identify specific features of those patterns. This is especially true if the patterns can develope over a variable number of bars.
-Bruce
Personal Criteria Formulas
TC2000 Support Articles
|
|
Registered User
Joined: 5/9/2011
Posts: 80
|
Bruce, the formulas below are for calculating the Linear Regression Slope values for periods of 83, 74, and 24... If we look specifically at the slope values of each equation, are the values of the slope equivalent to standard slope value calculations for standard linear lines that are typically expressed as y= mx + b where m is the slope? Thanks, Jim
(41.5 * C + 40.5 * C1 + 39.5 * C2 + 38.5 * C3 + 37.5 * C4 + 36.5 * C5 + 35.5 * C6 + 34.5 * C7 + 33.5 * C8 + 32.5 * C9 + 31.5 * C10 + 30.5 * C11 + 29.5 * C12 + 28.5 * C13 + 27.5 * C14 + 26.5 * C15 + 25.5 * C16 + 24.5 * C17 + 23.5 * C18 + 22.5 * C19 + 21.5 * C20 + 20.5 * C21 + 19.5 * C22 + 18.5 * C23 + 17.5 * C24 + 16.5 * C25 + 15.5 * C26 + 14.5 * C27 + 13.5 * C28 + 12.5 * C29 + 11.5 * C30 + 10.5 * C31 + 9.5 * C32 + 8.5 * C33 + 7.5 * C34 + 6.5 * C35 + 5.5 * C36 + 4.5 * C37 + 3.5 * C38 + 2.5 * C39 + 1.5 * C40 + .5 * C41 - .5 * C42 - 1.5 * C43 - 2.5 * C44 - 3.5 * C45 - 4.5 * C46 - 5.5 * C47 - 6.5 * C48 - 7.5 * C49 - 8.5 * C50 - 9.5 * C51 - 10.5 * C52 - 11.5 * C53 - 12.5 * C54 - 13.5 * C55 - 14.5 * C56 - 15.5 * C57 - 16.5 * C58 - 17.5 * C59 - 18.5 * C60 - 19.5 * C61 - 20.5 * C62 - 21.5 * C63 - 22.5 * C64 - 23.5 * C65 - 24.5 * C66 - 25.5 * C67 - 26.5 * C68 - 27.5 * C69 - 28.5 * C70 - 29.5 * C71 - 30.5 * C72 - 31.5 * C73 - 32.5 * C74 - 33.5 * C75 - 34.5 * C76 - 35.5 * C77 - 36.5 * C78 - 37.5 * C79 - 38.5 * C80 - 39.5 * C81 - 40.5 * C82 - 41.5 * C83) / 49385 <
(37 * C + 36 * C1 + 35 * C2 + 34 * C3 + 33 * C4 + 32 * C5 + 31 * C6 + 30 * C7 + 29 * C8 + 28 * C9 + 27 * C10 + 26 * C11 + 25 * C12 + 24 * C13 + 23 * C14 + 22 * C15 + 21 * C16 + 20 * C17 + 19 * C18 + 18 * C19 + 17 * C20 + 16 * C21 + 15 * C22 + 14 * C23 + 13 * C24 + 12 * C25 + 11 * C26 + 10 * C27 + 9 * C28 + 8 * C29 + 7 * C30 + 6 * C31 + 5 * C32 + 4 * C33 + 3 * C34 + 2 * C35 + C36 - C38 - 2 * C39 - 3 * C40 - 4 * C41 - 5 * C42 - 6 * C43 - 7 * C44 - 8 * C45 - 9 * C46 - 10 * C47 - 11 * C48 - 12 * C49 - 13 * C50 - 14 * C51 - 15 * C52 - 16 * C53 - 17 * C54 - 18 * C55 - 19 * C56 - 20 * C57 - 21 * C58 - 22 * C59 - 23 * C60 - 24 * C61 - 25 * C62 - 26 * C63 - 27 * C64 - 28 * C65 - 29 * C66 - 30 * C67 - 31 * C68 - 32 * C69 - 33 * C70 - 34 * C71 - 35 * C72 - 36 * C73 - 37 * C74) / 35150
AND
(37 * C + 36 * C1 + 35 * C2 + 34 * C3 + 33 * C4 + 32 * C5 + 31 * C6 + 30 * C7 + 29 * C8 + 28 * C9 + 27 * C10 + 26 * C11 + 25 * C12 + 24 * C13 + 23 * C14 + 22 * C15 + 21 * C16 + 20 * C17 + 19 * C18 + 18 * C19 + 17 * C20 + 16 * C21 + 15 * C22 + 14 * C23 + 13 * C24 + 12 * C25 + 11 * C26 + 10 * C27 + 9 * C28 + 8 * C29 + 7 * C30 + 6 * C31 + 5 * C32 + 4 * C33 + 3 * C34 + 2 * C35 + C36 - C38 - 2 * C39 - 3 * C40 - 4 * C41 - 5 * C42 - 6 * C43 - 7 * C44 - 8 * C45 - 9 * C46 - 10 * C47 - 11 * C48 - 12 * C49 - 13 * C50 - 14 * C51 - 15 * C52 - 16 * C53 - 17 * C54 - 18 * C55 - 19 * C56 - 20 * C57 - 21 * C58 - 22 * C59 - 23 * C60 - 24 * C61 - 25 * C62 - 26 * C63 - 27 * C64 - 28 * C65 - 29 * C66 - 30 * C67 - 31 * C68 - 32 * C69 - 33 * C70 - 34 * C71 - 35 * C72 - 36 * C73 - 37 * C74) / 35150 <
(12 * C + 11 * C1 + 10 * C2 + 9 * C3 + 8 * C4 + 7 * C5 + 6 * C6 + 5 * C7 + 4 * C8 + 3 * C9 + 2 * C10 + C11 - C13 - 2 * C14 - 3 * C15 - 4 * C16 - 5 * C17 - 6 * C18 - 7 * C19 - 8 * C20 - 9 * C21 - 10 * C22 - 11 * C23 - 12 * C24) / 1300
|
|
Worden Trainer
Joined: 10/7/2004
Posts: 65,138
|
Yes, the indivudal slope calculations within those formulas would be m in y = mx + b for periods of 84, 75 and 25.
-Bruce
Personal Criteria Formulas
TC2000 Support Articles
|
|
Registered User
Joined: 5/9/2011
Posts: 80
|
Interesting how the values returned when using LR as Indicator do not feel right relative to the visual. I realize scaling of the various axis will affect the numerical values vs the visual... Jim
|
|
Worden Trainer
Joined: 10/7/2004
Posts: 65,138
|
You can verify fairly easily that the formulas are correct. The endpoint of the following will line up with an 84 period linear regression line.
AVGC84 + 41.5 * (41.5 * C + 40.5 * C1 + 39.5 * C2 + 38.5 * C3 + 37.5 * C4 + 36.5 * C5 + 35.5 * C6 + 34.5 * C7 + 33.5 * C8 + 32.5 * C9 + 31.5 * C10 + 30.5 * C11 + 29.5 * C12 + 28.5 * C13 + 27.5 * C14 + 26.5 * C15 + 25.5 * C16 + 24.5 * C17 + 23.5 * C18 + 22.5 * C19 + 21.5 * C20 + 20.5 * C21 + 19.5 * C22 + 18.5 * C23 + 17.5 * C24 + 16.5 * C25 + 15.5 * C26 + 14.5 * C27 + 13.5 * C28 + 12.5 * C29 + 11.5 * C30 + 10.5 * C31 + 9.5 * C32 + 8.5 * C33 + 7.5 * C34 + 6.5 * C35 + 5.5 * C36 + 4.5 * C37 + 3.5 * C38 + 2.5 * C39 + 1.5 * C40 + .5 * C41 - .5 * C42 - 1.5 * C43 - 2.5 * C44 - 3.5 * C45 - 4.5 * C46 - 5.5 * C47 - 6.5 * C48 - 7.5 * C49 - 8.5 * C50 - 9.5 * C51 - 10.5 * C52 - 11.5 * C53 - 12.5 * C54 - 13.5 * C55 - 14.5 * C56 - 15.5 * C57 - 16.5 * C58 - 17.5 * C59 - 18.5 * C60 - 19.5 * C61 - 20.5 * C62 - 21.5 * C63 - 22.5 * C64 - 23.5 * C65 - 24.5 * C66 - 25.5 * C67 - 26.5 * C68 - 27.5 * C69 - 28.5 * C70 - 29.5 * C71 - 30.5 * C72 - 31.5 * C73 - 32.5 * C74 - 33.5 * C75 - 34.5 * C76 - 35.5 * C77 - 36.5 * C78 - 37.5 * C79 - 38.5 * C80 - 39.5 * C81 - 40.5 * C82 - 41.5 * C83) / 49385
The endpoint of the following will line up with a 75-period linear regression line.
AVGC75 + 37 * (37 * C + 36 * C1 + 35 * C2 + 34 * C3 + 33 * C4 + 32 * C5 + 31 * C6 + 30 * C7 + 29 * C8 + 28 * C9 + 27 * C10 + 26 * C11 + 25 * C12 + 24 * C13 + 23 * C14 + 22 * C15 + 21 * C16 + 20 * C17 + 19 * C18 + 18 * C19 + 17 * C20 + 16 * C21 + 15 * C22 + 14 * C23 + 13 * C24 + 12 * C25 + 11 * C26 + 10 * C27 + 9 * C28 + 8 * C29 + 7 * C30 + 6 * C31 + 5 * C32 + 4 * C33 + 3 * C34 + 2 * C35 + C36 - C38 - 2 * C39 - 3 * C40 - 4 * C41 - 5 * C42 - 6 * C43 - 7 * C44 - 8 * C45 - 9 * C46 - 10 * C47 - 11 * C48 - 12 * C49 - 13 * C50 - 14 * C51 - 15 * C52 - 16 * C53 - 17 * C54 - 18 * C55 - 19 * C56 - 20 * C57 - 21 * C58 - 22 * C59 - 23 * C60 - 24 * C61 - 25 * C62 - 26 * C63 - 27 * C64 - 28 * C65 - 29 * C66 - 30 * C67 - 31 * C68 - 32 * C69 - 33 * C70 - 34 * C71 - 35 * C72 - 36 * C73 - 37 * C74) / 35150
And the endpoint of the following will line up with a 25-period linear regression line.
AVGC25 + 12 * (12 * C + 11 * C1 + 10 * C2 + 9 * C3 + 8 * C4 + 7 * C5 + 6 * C6 + 5 * C7 + 4 * C8 + 3 * C9 + 2 * C10 + C11 - C13 - 2 * C14 - 3 * C15 - 4 * C16 - 5 * C17 - 6 * C18 - 7 * C19 - 8 * C20 - 9 * C21 - 10 * C22 - 11 * C23 - 12 * C24) / 1300
In these formulas y = b + mx where b is the mid-point of the linear regression line m is the slope an x is (period - 1) / 2.
-Bruce
Personal Criteria Formulas
TC2000 Support Articles
|
|
|
Guest-1 |