This is a power function regression of tops and bottoms. There is a separate regression for open, high, low, and close values for tops and bottoms.
There is a separate regression for each possible iteration of bottoms ( both bottoms of a double bottom, only the lowest bottoms of a double bottom, the first bottom of a double bottom, etc.)
The green circles show which data points were used for the regressions. On those days, all OHLC values were used to perform separate regressions.
All power function regressions returned R^2 values over .97, with the vast majority being >0.99
The upper regression lines accurately predicted both tops of the recent cycle. Note that those data points were NOT part of the regression, showing the tremendous predictive potential of this method.
The lower regression lines have a wide range due to the multiple iterations of regressions performed.
Despite the wide range of the bottom regression functions, none of them indicate that the current cycle has reached a bottom.
The equations for each regression can easily be viewed in the script.