Is Discounted Cash Flow Analysis Historically Testable?

I recently read this article which said that DCF is untestable because it depends on predicting future cash flows, which are pretty much unpredictable.
https://corpgov.law.harvard.edu/2022/04/20/the-dcf-valuation-methodology-is-untestable/

I’m curious, though, if anyone has tested DCF using historical data on cash flows to see if prices really do average around the DCF valuation (if you use accurate predicted cash flows). For instance, you could pretend to go back 5 years and perform a 5 year projected DCF on an equity using all of the real historical data, then compare your DCF answer with the price 5 years ago to see how accurate it was. Then, if you plotted the price and DCF over time, you might get an idea of how accurate DCF is for that equity. Do this for all equities, and you might learn what types of companies DCF is useful for.

I’m interested in this perspective for both the perpetuity and multiples DCF methods.

Is there any research on this? Am I misunderstanding DCF? Am I a dumb idiot? Undeniably. Thank you.