It's no secret that algorithmic trading is in vogue on Wall Street. But beyond the crush of numbers, there was always a human element to interpret news and make appropriate decisions. You may think a computer is incapable of "reading" an article, say, on new guidance from a company and then making good trading decisions, but a group of vendors think you are wrong--and they are proving it. For example, Reuters launched a service this month that scans news articles and then assesses a positive or negative rating, which triggers an appropriate buy, sell or other order. It works by using linguistics software to provide numerical sentiment scores to the text based on words and phrases. We'll likely see others follow suit. This is blackbox trading in the extreme. If you can get the program to work, it seems like a good idea. But what happens when the machine fights back? 

For more:
- here's the Financial News Online article