What's appalling is how even a cursory glance at the basic techniques of today's "Big Data" numerical financial analysis paper over deep and intractable issues when considering long-term investments. Among many examples of the dead-end canals lurking for those picking stocks purely by the numbers, the most shocking to me is something called sentiment analysis: You know, this Big Data trick in which Web content -- usually in the form of Twitter tweets or Facebook ( FB) posts -- are agglomerated and analyzed to get a feel for how people feel about the world around them. Sentiment analysis is what got fooled when The Associated Press had it's Twitter account hacked and a fake White House bombing sent markets into a tailspin. Take a look at Sentiment140, or Opinioncrawl.com if you seek a direct sense for this family of tools. But when I actually tried my hand at creating just such a sentiment analysis a few weeks back, as interesting as it was to capture two hours of live tweets and use a coding language called Python to sniff out the sentiment of those tweets, a basic and obvious question lurks: How many tweets does one need for these mostly meaningless 140 character phrases to magically morph into a meaningful "sentiment" an investor can trade against? Big data practitioners assume the logic of statistics is at work here. More numbers, more tweets and more data make a more accurate model. It only took one fake tweet to crash the Dow, though. So who really knows how many of these suckers you need. It all creates a level of uncertainty confirmed by data analysis experts. If practitioners do not understand how these tools work and trust in the black box of analysis techniques, you can wind up in real trouble -- that's how Bill Howe, director of research for scalable data analytics at the University of Washington, described it in a recent lecture at Coursera, the online education service offering excellent training in data analysis. Buy and hold what's real
And that, friends, makes Gardin's simple, elegant and powerful approach to his photos such a flash of investing genius. In this sinking lagoon of data overload and confusion, when it comes to finding investments worth owning over time, borrowing the nondigital approach that works for Gardin should work, in an adapted form, for investors. The world does not live entirely by numbers. It's out in the world itself. So go out and find something that interests you and makes sense to you. Then figure out a simple way to invest in that interest. And let reality do the rest. Because hoping numbers will do your work for you is nothing more than a gondola ride to nowhere.