Standard Deviations from a Linear Regression Trendline?

Question

How do you calculate the value that is 2 Standard Deviations from a Trendline. For instance, you have 1000 daily data points. You calculate the value of a Linear Regression trendline over the last 100 days from you daily data points. Do you calculate the 2 Standard Deviations of your underlying data points or 2 Standard Deviations for the Linear Regression trendline itself when plotting prediction bands around your trendline. Also, what about Standard Error bands? Are they calculated using the underlying data or the trendline values? Finally, if you mulitply a Standard Error over 100 days by its corresponding 95% value from the Student's T-Distribution Table, will that give you an accurate value for prediction bands around your Linear Regression trendline?

Answer 1

This must be the fifth time you ask the same question.

At least two people, including myself, gave you an answer already.

I gave it here:

https://answersrip.com/question/index?qid=20061124000838AAVXOMl



But the important thing is my remark in point number four....

Answer 2

utilising linear regression, it does not incredibly make experience to declare that the trendline "has a cost of one hundred fifteen". i choose to propose staring at an internet site that explains linear regression in some element. even nevertheless, the line is expressed as y = a + bx the place y is the based variables, x are the autonomous variable, and a and b are the parameters of the outfitted equation. This incredibly gets too complicated to describe intimately right here, so i choose to propose which you inspect some specific information on linear regression to get the respond.

Answer 3

the error would be based on your trend line. as you are using that to forecast beyond the end of your data. and you kind of eliminate your error bands or anything like that by doing the trend line or are you not using OLS. and i dont think multiplying your STD will do much besides give you a bigger number. the STD is a function of the variance over the line so multipling it will not do much for you. but do you really need STD off the trend line or off the bell curve distribution. if you need it off the trend line you use the trend line as a mean for the distibutions and create it off that then you have a one two and three STD lines. been so long since i did it by hand not sure how its done on paper.