There doesn't appear to be any period in question where you could invest in a broad-based index of funds and withdraw that investment 20 years later at a loss. That's what is shown in the FA and what the comment above means. Nobody's saying it has never declined.

What isn't true about the above statement? It's incredibly specific, yes, but it shows that at least historically buying and holding over time limits huge gains but also limits losses.

It doesn't limit anything. Let's say you put 100k in the market today. With probability p you will have 8% return at some unknown point in time in the future (your argument may be false even from a simple temporal point of view, take a day in which the market plummeted, and take the closing price of that day, not the end of the year value, and 20 years prior I doubt you would have made a profit, but I can't make any statistics as of now). That means your 8% is unknown. So you haven't really mitigated any loss. The distribution of returns is unknown, at any time period, dayly, yearly, whatever, so you can't know your p, so any wishful calculation is nothing.

Now, let's assume what I said up to now is wrong. I don't pretend to know the absolute truth anyway. Reflect on this: even if something has made a profit buying and holding for 20 years consistently for a millennia, it means nothing for my future. What guarantees me that I will make my profit in my 20 years? I can't travel back in time. Before discovering black swans, everyone thought swans could only be white.

Also, 150 years of data on a phenomenon that changes almost daily is nothing.

The linked page covers this in detail. The market goes down often, and sometimes goes down significantly for several years!

The specific point being made is that there has never been a 20-year period where the U.S. stock market declined -- when comparing the start versus end value of the S&P 500 index, on a dividend / inflation adjusted basis.

it's not an average. its saying you buy at the start of the 20 year period and withdraw after 20 years.

The averaging comes from zooming in and zooming out, which apparently is some kind of scientific method I am not aware of.

not sure what "zooming" you're referring to.