One justification for bonds is their low correlation to stock performance. Bond funds can even have a slight negative correlation according to portfolio visualizer's data. So when stocks drop, it's very likely bonds have gone up slightly. Note that doesn't apply to high yield "junk" bonds, which tend to move with stocks more.
People think that bonds and stocks are negatively correlated which is why they hold a mix of both, but this is incorrect. They are POSITIVELY correlated over the long term, and the correlation increases the longer the holding period. That is why stock/bond portfolios have worked so well for so long in the disinflationary environment we've had since the 1980s, falling yields have driven bond prices higher, while the falling interest rates have provided a strong tailwind to equities (and real estate too).
Market data from portfolio visualizer shows a negative correlation between stocks and bonds. It's also conventional wisdom that stocks and bonds have low correlations. Your post claims that market data and conventional wisdom are incorrect, but you fail to provide evidence for your views.
https://www.portfoliovisualizer.com/asset-class-correlations
For those who rebalance annually, change the time period on that page to "annual", and notice how AGG (iShares Aggregate Bond ETF) has negative correlations with the S&P 500, mid-caps and small-caps.
It is historically correct that bonds were usually counter-correlated with stocks, because interest rates were cut when things occurred that would hurt stocks, like recessions.
Since the early 80's however, both stocks and bonds have gone up together as rates were cut from historically high levels - around 15% for treasuries and even higher for corporates! That particular scenario is not happening again anytime soon.
Looking a few years into the future, there is little remaining room to cut interest rates below zero, and one could argue lots of pressure building to raise them. If something happens that causes stocks to go down, there will not necessarily be the traditional rate cut to boost bonds. Fed officials have already expressed resistance to negative rates like happened in the Eurozone. Instead, we'll probably see more QE or helicopter money to address the next recession because these worked like a charm in 2020 and prevented a depression, even near ZIRP. And if that doesn't happen (e.g. because of austerity politics) then stocks could tank without any benefit to bondholders.
Scenarios:
1) Recession addressed with QE/heli-money instead of interest rates: bonds remain flat while yielding nearly nothing. Stocks zig zag.
2) 3-5% inflation which the fed tries to control with QT instead of interest rate increases (have you seen the size of their QT war chest?). Stocks would probably decline while bond investors get large negative real returns and rates would eventually rise anyway.
3) Fast growth / low inflation allows the fed to keep rates low for years: bonds flat and yield nearly nothing while stocks boom.
4) 20% collapse in value of USD dollar vs. basket. Fed attempts QE to keep interest rates artificially low but market demand causes interest rates to rise. Bonds lose piles of money as interest rates go up to India levels.
The one bonds-up/stocks-down scenario where an investor would get a sustained opportunity to rebalance from very appreciated bonds to very cheap stocks would be a bear market accompanied by a drop to sub-zero interest rates. If a deflationary crisis occurred and the 10y US treasury went from today's 1% to -0.5% like in Germany, bonds would be going up while stocks go down. Holding bonds yielding less than inflation is a bet on this one scenario. In the other scenarios, depending on the specifics, either cash or stocks would probably do better. Regardless of the realistic probability one assigns to each scenario, the expected value of bonds is low enough to be virtually equal to cash. E.g. the probability weighted outcome of all the scenarios has to be close to an inflation-adjusted zero, unless one is a zealous about a particular scenario.