<<I must be watching almost the same things as you. Seems like when one finally finds a pattern that works...then it doesn't any more.>>
Nothing works all of the time. If Marc has done adequate backtesting on his methodology and historically he finds that his favored process successfully returns profits 7 or 8 times out of 10, then he just needs to ride out the few times that it is not profitable.
BTW, I went to cash today from 40% long. My model achieved the "go to cash/optional short" threshold for the first time since early September.