In http://www.financial-hacker.com/seventeen-popular-trade-strategies-that-i-dont-really-understand/ jcl stated clearly that it is neither proven nor disproved that traditional TA works. Many TA instruments are dealing with trend recognition, trend following and recognition of trend reversals. Often I read that the quantity of instruments is an indication that TA won't work, because otherwise one or two with predictive power would be enough.
I'm not sure but it seems to me that there is often a discrepancy between the usage of TA's indicators and the way they should be used. Consulting the TA 'bible' (J.J. Murphy, Technical Analysis of the Financial Markets) I read there more about a concept where indicators' results are used as puzzle pieces that if cleverly choosen, interpreted and arranged can constitute an image of the present markets' situations and with more or less probalility can also make forecasts of further developments. Murphy describes the importance of indicators that deal with one intrument or market as well as the importance of indicators dealing with intermarket observations or secondary paramters like e.g. the volume and the use of those instruments in different time frames to determine the actual trend situations in markets which are additionally interdependent.
I do not read there that any indicator or any fixed combination of indicators based on one intrument alone can reliable predict its future price curve. Therefore the fails of such tries should not be very amazing. To cut it short - in the field of using technical analysis and therewith analysing trends there is IMO still very much room for improvement!