Computers to Trade Stocks Based on News Reports

Stock markets and mutual funds are already ruled by computers that watch the market diligently and decide when to buy and sell based on fluctuations in stock prices and other market conditions. That's yesterday's news. Tomorrow, computers will be watching the news wires, trading based on whether news about a company is good (Sales are up!) or bad (CEO caught with a prostitute!).

Reuters is launching a service today that tags its news feed with machine-readable keywords, to the tune of 8,000 news stories a day. Observers expect other news services to follow suit... which is where the problems could begin.

For 99 percent of trades, this news service will probably work well. But what happens when tags are wrong (and I know you've seen as many stories that begin with "CORRECTION")? While computers can read these tags, it's still up to error-prone humans to create them.

When other companies get into the game, things might totally fall apart. I'm counting the days until I read a (serious) headline that reads, "Mutual fund loses millions after selling stock based on a report from satire news site the Onion."

Still, news-based trading is probably just a matter of time. As the above linked story notes, computers currently trade 60 percent of all stocks in the United States, and that figure is certainly not headed down. But when it's computer vs. computer, every tiny advantage is critical, since there's no intuition or human analysis behind the buying and selling.

What do you think? Would you entrust your cash to a system like this? (Hint: You may not even know that you're already doing it.)