I think this new WIP version should be ok. Because code is read from top to bottom, sometimes counting were wrong, so I stacked the code in another way (my bad sorry).

About the Close superior or inferior to the one 4 bars earlier, it is not how it is coded in the original code posted by joerie1, it is looking for the current Close to be superior to highest one of the last 3 bars (up swing) and vice versa for the down swing, but I modified as per your suggestion..