Our history is nine years of crypto and it is our own data. Every choice we ever made was made while looking at it. That is a real limit and no amount of internal testing fixes it.
So we went looking for a series with the same shape and none of our fingerprints on it, and settled on a covered-call index on US equities going back to 1988. A different asset class, a different era, a different set of participants. Structurally the same payoff: give up some of the upside, get paid for it, take the tail when it comes.
We compared the risk characteristics rather than the returns, since the returns are not comparable and were never the point.
They agreed closely. The comparison is in Letter 001.
Two consequences. What we do is a property of the payoff class rather than a quirk of our period, which is reassuring. And the flip side comes with it: that index has periods of lagging the market measured in years, not months, and there is no reason we would be exempt. We took both halves of that finding, because taking only the first would have been the point at which this stopped being research.