I think that there are a couple things at play here.
On the one hand, TA always works. Markets almost always behave and react in a similar and recognizable pattern. (e.g. rally, break, consolidation, chop / rally / break....).
But, there is more going on and so you can't just say that a certain momentum, relative strength, or setup is truly consistently predictive (it's an over simplification). There are always going to be exceptions to the rule.
One thing that is going on is that traders (and robot traders) are using TA on spreads which are the risk adjusted differentials of things like indexes. So you can have a rally going on in an index spread that is pushing the price action in the stock averages (contrary to classical TA). The sectors are being traded against each other, and equities are always being traded against the bonds.
At the end of the day, TA and setups / indicators / screening tools are just ways to control risk/reward outcome. In the modern era, professional risk managers are using much more. They are building long/short books across global equity markets and bond duration.
There are many ways to control risk, TA is just one of the most basic. You have to remember, when you are trading billions of dollars, the market just isn't liquid enough to time your entry and exit. The liquidity just isn't there in such a short time frame.
That's why you see so much action in the price. There could be 8 hours of consolidation and then a break and massive rally afterward because the entire liquidity situation has changed!
Here is a great post about the new markets from a REAL PRO TRADER.
http://tradestrongmanagement.blogspot.com/