Been busy this past week looking at fractals, identifying them from historical price data and trying to find relevance in them. So far I’ve created a web interface that can spot fractals using a correlation window with various input parameters.
The tool can spot similar movements across a period, especially sharp movements. I’ve found that by using a weighted correlation window that favors the shorter windows (more recent prices) to be an improvement over simply comparing one window period to another.
Yesterday, I manually scanned some stocks and came across $WB which had strong signals for an up day. While this came true, I also tracked other stocks and plan to refine the filtering process and to create a script to automatically scan stocks that follow certain criteria.
Enjoy the screenshot grab below of v0000001…
Personal notes: corr_sum ≥ 0.85, perc_change ≥ 0.5%, 5 of 5 top corr in the same direction