A comparison with the LRMA (red line) on the same chart. Bolinger bands are disabled for demonstrativeness.
It can be seen that with the same size of selection (30) ParMA has a lesser delay comparing to the LRMA
with the comparable jumps on the price chart.
A notice to the code: properly, the calculation of the parabolic regression was taken from the book:
and left without changes not to embarrass the code.