The 50/200 crossover is the first system most people code, and the first one they abandon. On daily index data since 1990 it trades about once every two years, which means any single decade tells you very little about its edge.
What the rule does well is keep you out of long drawdowns. What it does badly is everything else: it re-enters late, it whipsaws in sideways years, and transaction costs matter more than the textbooks admit once you move to single stocks.
A fairer test is to compare it against holding the same asset with the same average exposure. Measured that way, most of the apparent outperformance is simply lower volatility, not extra return.