Statistical significance of TA-patterns (S/R, Chart & Candlesticks)

Hi, I've always wondering if there are any tests made for testing the Statistical significance of all these Technical analysis patterns. I dont want to propagate the efficient market hypothesis as this is not my intention.

For Support / Resistance in example, it would be a benefit to know which levels are actually important. There are historical resistances, pivots, fibs, round numbers, some are even using halfs, and quarters...and if you all draw them in your chart, there are plenty of lines.

In FF.com you find statements like "xxx works" or "i find xxx significant" but its always that mystic blabla, not backed up by "hard facts" aka statistics.

It would be great if somebody can point me to statistics like that, or better tell me how to perform such tests by myself as I am unfornutaly not a statistic pro :/

thanks

There is nothing, absolutely nothing that can give you an edge
If there was one there would be no risk. You cannot remove risk, never ever, therefore to find that something that is statistiscally in your favor, simply does not exist.

In that case there would be no profitable traders ever. Just having profitable traders means they have a statistical edge with their system. And there are statistics on certain patterns and what their historical edge is, its just not stuff that you will find on the internet for free.

http://www.ny.frb.org/research/epr/00v06n2/0007osle.pdf

There're many ways to trade S/R, the question is whether what you do works not whether S/R works...

Bootstrapping http://www.automated-trading-system.com/bootstrap-test/