Much like index investing, this is a theoretically justified approach based on some basic assumptions (almost uncorrelated assets and an almost efficient market). The fact that it does well in back tests should be seen as additional evidence in its favor I think.
The back-testing graphs and OP posts near it are very telling. I disagree that this is "like index investing". This is a very active bet on interest rates.
Index investing is based on a logic and math:
Let M be the total market of stocks
Let P be the proportion of the stock market held by passive investors
Let A be the stocks held by active investors.
The market is composed of passive and active investors.
M = PM + A
By definition, A = (1-P)M, so in the aggregate, active investors own the exact same ratio of stocks as passive investors.
Paying lower fees is better, so in the aggregate passive investors will be better off.
A corollary is that active investors can only get ahead by taking from other active investors. In order to get ahead of passive investors, they have to consistently over-perform the market by more than the fees. Empirical evidence shows this (consistent over-performance) is extremely rare.
You are making an assumption of low correlation between stocks and bonds. However,
I'm not sure that's a good assumption. Specifically from that article: "Stock-bond yield correlations have been largely positive since the late 1990s, rose strongly during the global financial crisis...", which means
right when someone wanted it most, the correlation did the worst possible thing it could.
In reality, what this suggests is much more along the lines of what Long Term Capital Management did. They had "a theoretical model of what the relationships between different but closely related fixed income securities should be" (
quoting the Wikipedia article), but it got worse (bankrupting them due to leverage) before it returned to their expected trend.
What's worse, is that the process clearly produces horrible results in rising interest rate environments. The OP of that thread admits it right after the grand backtest. If this really is just a major bet on interest rates, why not bet more directly (and likely for less fees)? [which really is: if you can really predict interest rates, why are you posting here and not rich and retired?]