But isn't it true that IF a method is based on a market phenomemon with is relatively timeless, the method if based on indictors can survive if the parameters of the indicators are adjusted to the changing environment. ( and in sorm form it is discretionary then :-) )

For example, trendfollowing and mean reversion strategies are very old and still working, yet the form they take changes over time due to new market participants and conditions.

So if a method with indicators is based on such a phenomenon it can survive, if the user adapts the parameters to the environment. When a method is based on some arbitratry collection of indicators which do not support/measure a good concept, the method will likely fail. At least, this is what I think within my limited experience.