Linear Regression Slope: the gradient of the least-squares line fitted over the
last periods closes, in price per bar.
It asks a different question from LSMA, which plots where that same line ends.
Dropping the level and keeping only the gradient is what lets the study say
that an advance is still an advance but no longer as steep, at a bar where
price itself is making a new high. Zero is the whole reading: above it the fit
rises, below it the fit falls.
The x axis is unit-spaced and the reference scales the result by nothing, so
the answer is per bar rather than per window. Only that spacing matters: a
least-squares slope is unchanged by shifting x, so running x from 1 to
periods (the reference's own convention, newest bar highest) and running it
from 0 give the identical number.
Linear Regression Slope: the gradient of the least-squares line fitted over the last
periodscloses, in price per bar.It asks a different question from LSMA, which plots where that same line ends. Dropping the level and keeping only the gradient is what lets the study say that an advance is still an advance but no longer as steep, at a bar where price itself is making a new high. Zero is the whole reading: above it the fit rises, below it the fit falls.
The x axis is unit-spaced and the reference scales the result by nothing, so the answer is per bar rather than per window. Only that spacing matters: a least-squares slope is unchanged by shifting x, so running x from 1 to
periods(the reference's own convention, newest bar highest) and running it from 0 give the identical number.