In Financial News Circles They Say You Should Buy On The Rumor And Sell On The Fact And Now There Are Computer Programs That Can Read That Information And Do Those Trades Automatically

Philip Stone June 27, 2007

Don’t be fooled into thinking that by reading Reuters or Bloomberg on the Internet that you know what is going on in the financial world as quickly as financial institutions around the world. The financial people pay hundreds of millions of their currencies globally to buy their news directly from information providers so they can get the market breaking news first and make their trades ahead of the competition.

Around the world, whether you’re in Tokyo; London, or New York, that same news will appear within a split second on financial trading screens and then it is up to the human being to read, digest, and make a decision whether to trade. And the time it takes to make that decision could be difference in being the first into the market or being a runner-up, probably with less profit.

So the obvious solution here is to get rid of the human intervention and get machines to read the news copy, make the decision, and make the trade since the machines can do that far faster than humans. Science fiction into the future? No, it’s here today.

Most information vendors are now deeply into providing machine-readable news. Reuters, for instance, recently bought ClearForest Ltd., a provider of text analytics solutions. "The acquisition will enable Reuters customers to efficiently and profitably search the wealth of news, financial and other relevant information available today, as well as to derive previously unavailable business insights from that content," according to Barak Pridor, ClearForest CEO.

ClearForest’s software allows unstructured text and enterprise data to be analyzed simultaneously, in effect giving a whole new meaning to business intelligence by taking large volumes of information and boiling it down into the bare facts needed to make decisions.

Reuters has also expanded its machine readable news services by including news analysis software from Corpora plc. That software assigns numerical “sentiment scores” to words and phrases found in stories and then it presents an overall positive, neutral or negative score to the company in the news story.  Add the scores from several stories and not just a company, but its financial sector, an index and the like can all be used to assess global market sentiment.

The magic words today in financial news reporting are “machine readable news” closely followed by “algorithms”. Programs now search for market-sensitive terms and computers are programmed to analyze that information and trade on such news within milliseconds of having received it.

A decade ago this writer was the Reuters news products manager for Europe. Middle East and Africa. In those days many financial traders made their decisions  based on just the one-line alerts of a few words on a breaking news story. All that is really going on now is a refinement of that system to take out the “slow” human and replace he or she with a computer that has “what if” decisions already programmed into it.

Now, of course, this is not great news for all those financial journalists out there who are mighty proud of their 750 word take-outs on various subjects. Those stories are great for media customers but for the financial customers they are a luxury to be enjoyed when times are quiet. For the traders, the more one-line instant alerts the better. They are the primary impetus for a trade; traders are not waiting around to get the full story from any great journalistic verbiage.

In today’s world, in fact, the financial institutions are just as happy if the news suppliers would do away with sentences, let alone paragraphs. Just provide a computer-friendly string of words with the necessary numbers and the computer will figure it all out and make the trading decision in milliseconds.

Some financial traders, of course, are more cautious than others and while delving into such automation they are not completely sold that there shouldn’t be some sort of human oversight. But if the completely automated systems consistently get into the market before those with human oversight then that will be a nice to have that will likely disappear.

And this type of technology is not just limited to text. An Intel demonstration last week showed that given a mighty powerful processor similar benefits can be brought into the home and used, for instance, to automatically reduce taped video down to its most important points.

Research developed by Intel and Tsinghua University in China, showed that a two-hour taped football (soccer) match can be automatically reduced down to just five minutes of highlights. An eight-core Intel processor capable of 100 gigaflops identified key action points – goals, corner kicks, fouls and the like – and eliminated all the unexciting stuff.

If advertisers aren’t happy today with digital users zipping through their commercials just imagine how they will react to equipment that can be programmed to automatically make those commercials disappear!

Computers are already being used to generate set news stories – company earnings, economic statistics and the like -- and those stories flow automatically into the computers at financial institutions. Just more evidence that the time between a news organization having information and their clients getting that information is ever being reduced.

Some 40 years ago a book written about the American news agency United Press International was published under the title, Deadline Every Minute. In today’s world it would have to be renamed, Deadline Every Millisecond.

ftm Follow Up & Comments