I wonder if the crossover strategy would still outperform (on a Sharpe Ratio basis) starting in 2005 instead of 1998.
I also wonder if the crossover strategy would fail to outperform (on a Sharpe Ratio basis) once transaction fees were considered.
E.g. a major player rebalancing a position created distortions that allowed much smaller players to profit, but no distortion on the scale that would allow others trading at major volume to profit
It's an area of active research:
> Students of financial economics have largely attributed the appearance of momentum to cognitive biases, which belong in the realm of behavioral economics. The explanation is that investors are irrational,[4][5] in that they underreact to new information by failing to incorporate news in their transaction prices. However, much as in the case of price bubbles, other research has argued that momentum can be observed even with perfectly rational traders.[6]
* https://en.wikipedia.org/wiki/Momentum_(finance)
* https://en.wikipedia.org/wiki/Carhart_four-factor_model
Though the market, size, and value factors appear to explain >90% of returns (at least using US data):