Some additional observations prompted by Ramc's excellent comments:

Allocate Smartly is aware of the confounding effect of the long bull market in bonds. They identify strategies that are particularly sensitive to interest rates, which in such circumstances are likely to produce better historical they they will going forward from today. In addition to calling out the risk, they offer what they call "dynamic bond" variants of strategies that rely exclusively on intermediate treasuries as a risk-off asset.

They are also aware that backtests that seek to optimize over a long historical period incorporate a crystal ball effect, by which choices that matter to the early part of the period are shaped by conditions at the end. They have attempted to minimize this by developing what they call "meta" strategies: combinations of allocation strategies that are re-constructed year by year, using only data that would have been available at the time.

Finally, while their standardized report of results use trades on the last day of each month -- a necessity given how common that practice is in academic and other research -- they provide backtested results for every day of the month, to expose the variance this introduces. It is thus possible, with a little work, to identify the best, worst, and median trading days for any strategy, in order to get a more granular view of its results.

Baltassar