Anybody can produce good historical performance if the system is optimized enough. How do we know that it will perform just as well in the future?

The fact that the systems performed so well on a portfolio of  3000 stocks (Russell 3000 components) for 10 years suggests that the system is robust enough to perform well in the future. Of course long systems won’t perform so well in a bear market and the reverse for Short Systems. Fell free to request performance results for a different stock list or dates.