Happy Monday Bruce,

Over the weekend I used the PCF templates to create Linear Regression Slope and R-Squared indicators for 21-periods. Easy peasy.

But, ...

I prefer to plot Lin Reg Slope as a histogram and I really love the way you are able to use four crayons. 

Can you pull out your Crayolas on a Lin Reg Slope 21 Histogram?

What colors are you looking for and under what circumstances?

Bruce,

I can change the colors after you're done, though I'm sure your choices will be just fine. :)

Sorry, the usual circumstances ...

• Below 0 and Falling
• Below 0 and Rising
• Above 0 and Rising
• Above 0 and Falling

Thanks for your help!

Start by plotting a Custom PCF Indicator using the Histogram Plot Style and your formula for 21 period linear regression slope.

3 * (FAVGC21 - AVGC21) / 10

Choose Negative Value Color and choose the values for the "rising" colors.

Now overlay another Custom PCF Indicator into the same pane and scale as the first Custom PCF Indicator (make sure to choose & Scale With).

You will select the Histogram Plot Style and Negative Value Color for this Custom PCF Indicator as well (using "falling" colors), but use a formula which only plots when the linear regression slope is decreasing.

IIF(FAVGC21 - AVGC21 < FAVGC21.1 - AVGC21.1, 3 * (FAVGC21 - AVGC21) / 10, 1/ 0)

Bruce, will give it a shot. Thanks!

Linear Regression, Slope of has always confused me. I would expect slope to predict rate of change or the steeper the slope, the higher the price appreciation.

On a chart today, the Lin Reg 21 slope of ISRG is 1.18 using the formula above If I drag my cursor from the low point of the LR line to the high point, TC2000 tells me that the annualized  appreciation is ~76%

When I do the same thing for ULTI with a lower LR Slope of 0.90, the annualized appreciation is ~123%

For FFIV, slope is 1.62, but anualized appreciation is ~211%

I seem to be missing something (not the first time, nor the last, sadly) Can you help me understand the value of the LR Slope, or how I should be interpreting it?

Thanks

A linear regression line attempts to a draw a straight line which best fits the data through which it is drawn.

It uses the "least squares" method to do the calculation and at least with most of the formulas posted, the calculations are done using an arithmetic scale.

The slope is the net change from one bar to the next along linear regression line. The net change is the same from one bar to the next so the line is straight in arithmetic scaling and curved in logarithmic scaling.

The calculations for the slope are from one bar to the next and not the net change along the entire length of the linear regerssion line.

The calculations are not percentage based, so the same slope on a low priced stock is going to represent a greater percent change than on a higher priced stock.

If I were to divide the Slope by Price, would the result allow me to rank appreciation of the Linear Regression line?

(A few quick eyeballs would indicate thus)

Thanks

Yes, that should work. And if you multiply it by 100, you would even get a sort of percentage (although it wouldn't actually be the same percentage from one bar to the next). I would also probably divide the slope by a simple moving average of the same period as the linear regression slope instead of just using price.

Thanks. I had worked out the multiply by 100 thing, but using the average close is an improvement

You're welcome.