by Irfan Khan IT World Blog
If you spend any time following front-page news you know that most of it is grim. Floods. Wars. Earthquakes. Robbery. Murder. It’s enough to make a reader skip to the sports section.
For savvy investors, though, the news is not just about tragedy; it’s about money. Or, more precisely, keeping an eye on the news is about identifying risk before its ramifications hit your investments.
That’s why I was intrigued by work done at the Yale Center for Analytical Sciences by researchers William Casey King and Michael Kane. They analyzed unstructured data from news feeds to see how it can affect risk in corporate bonds.
In their study King and Kane note that corporate bonds are widely considered safer investments than equities. As such, the market for these bonds is huge, more than $6.3 trillion, greater than the combined investments in treasuries or municipal bonds. However, as they learned, even a “safe” investment is not immune to the downside of very bad news.
The Yale professors analyzed the impact of the Transocean drilling platform catastrophe in the Gulf of Mexico in 2010 on BP’s bond prices. Not surprisingly, they discovered that BP’s share prices tumbled after news of the disaster became widespread. But, after a lag, the bond prices fell in line with the share price.
I do not pretend to be a financial wizard, so I won’t attempt to explain how mixed-asset hedging works. But according to King and Kane, had BP bond investors been using real-time, Complex-Event Processing technology to analyze the news about the disaster, they could have adjusted their hedge in bonds quickly enough to have avoided substantial losses compared with the more traditional hedge or using treasuries.
My point, however, is simple. There are bottom-line reasons for applying analytical tools to unstructured data even in real time. It’s not pie-in-the-sky theory. So, if your business gets rocked by news events or your brand is being battered on social networks, you don’t need to apply resources after the fact to clean up the mess. Instead, you can apply predictive analytics on the unstructured information to avoid the mess entirely, or, worst case, hedge its effect on your bottom line.
DCL: These Yale professors’ slide presentation (see the referenced study) appears to be an unabashed advertisement for Sybase, which would normally mean we would not discuss it here. But it also contains a lot of food for thought about complex event processing and the use of real-time event feeds in Business analytics. Many other commercial CEP tools currently on the market could have been used to do similar studies, as the reader will no doubt conclude.