Check out The Last Chartist (via TRB). It’s about the robots relegating charts to the role of confirmation rather than prediction. While I think it’s premature to declare the death of technical analysis, it has been smelling funny for many years, long before HFT. For more than 20 years computers and greater information dispersal have resulted in rapid, widespread implementation of anything that works even a little, which just as rapidly leads to it being rendered nearly useless. Trading shorter time frames, which has also become more popular over that time, also makes HFT and other computerized competition more brutal. Short-term data has always had a low information, high noise content and HFT has only amplified that noise, naturally making short-term price patterns less reliable.
What is a hapless human to do?
The easiest solution for filtering HFT and other noise is to look at longer time frames. Weekly and monthly charts look no different now than they did a century ago. Just because real-time data is readily available doesn’t mean you need to be using it.
Another response is to use indicators that are less easily manipulated by the actions of other market participants. Breadth indicators are some of my favorites for that reason, especially when calculated using only common stocks.
But the best response is to find an unexploited niche to make your own. If everyone, man or machine, has gone short-term the best response is to go long. If everyone is watching price, watch the people watching price or watch something else entirely. The fewer people playing your game the easier it is for you to win, and in the markets there’s no shame in winning by being the only one in the game.