The author proposes an investment strategy that would last 100 years without any changes.
His underlying assumption is that "we don't know anything about the future — only about the past" and therefore one should look to the past, across as many years and asset classes as possible — to construct a portfolio that would perform well based on past data assuming the future plays out in a similar manner.
This is a fascinating paper, and a wonderful question to pose — far better than the more basic questions which sit beneath most ETFs (e.g. how can we own the whole market? e.g. which segments or sectors are have what we want?)
He rightly states that most strategies are woefully ignorant of their strong recency bias, and many investors are effectively long (and likely leveraged) to the business cycle e.g. someone buying a home with 20% down is making a 5x leveraged bet on the business cycle as the value of their home, their job/income that pays for the home, and their stock and bond retirement portfolios are strongly correlated to the business cycle.
I agree with the approach of not assuming the future will be a repeat of the prior 3-4 decades. I disagree that the antidote is to look back even further. This is wiser than ignoring recency bias, but ignores anything we can know from first-principles about differences in the world today vs. say 100 years ago.
Human nature has not changed. Government incentives have not changed. Fiscal and monetary incentives have not changed. But businesses and assets have. The world's wisest businesses 100 years ago were winners — as are the wisest businesses today. Business models have evolved enormously (e.g. only in the recent few decades were zero marginal cost businesses possible). The investment community has evolved enormously (e.g. there is far more overall maturity in the market, as measured by the proportion of institutional money in the market, the proportion of passive and therefore predictable money in the market...a great alternative for those that would otherwise gamble, etc.) This is NOT to suggest that we should expect calmer seas ahead. Greed and fear still rule humanity. But it IS to suggest that learning from 100 years of market data is not the best approach — this market data could not teach us the best ways to structure a portfolio of wonderful businesses today. Most of those businesses didn't exist several decades ago, and most of the instruments one might use to manage their risk didn't either (e.g. gold was good...Bitcoin might or might not be better, but historical data can't tell us this).
