It doesn't quite matter how long the years of data are, it is still one single timeseries. That's one of the most important things in time series econometrics.

It doesn't quite matter how long the years of data are, it is still one single timeseries. That's one of the most important things in time series econometrics. You need to test long time series along multiple securities preferably across different geographies. Taking a 50/200 MA crossover over a single ETF literally proves nothing.