If it's hard to test realistic trading systems (long, short, flat), then that doesn't speak well of the usefulness of the Aronson testing methodology. The always-in systems he did test just have no relation to real-world trading, which makes me wonder what was his point?
Funny how none of this...