True but I'd question why a code set that did so well over such varied conditions would drop to a 10-16% level for 2 years. 20 years of test covers a huge range of market types, should be more than sufficient.
From my own experience I'm reasonably sure there is something to be found in those years. In my own program I call it a 'bogeyman'.
You CAN time the market!
But only after you trash the big 3 myths.
Myth 1 - its possible to get every point of every move ...Myth 2 - you'll never have a drawdown ......Myth 3 - you'll never have a loss