You have to understand that there is absolutely no way to 100% guarantee a turning point based on any type of signal analysis (he mentioned hurst waves, which is the high school version of fourier series). It's also important to quantify how far the turning point is expected to stretch; one could have a legitimate, yet, non profitable turning point, say in a flag type oscillatory pattern.
What would be more significant, IMO, would be to have a detailed table of how many times it worked and how many times it didn't. Always be cautious of cherry picked black box examples (especially with the incredible! description).
That being said, a rare intelligent discussion of these ideas was discussed right here on the optimized trend trading threads.