Monday, June 15, 2009

Mining the Twittersphere

Twitter or a Twitteresque technology will, in the not too distant future, emerge as one of the most disruptive Web 2.0 technologies to be ever created. Why? Because Twitter enables one to instantaneously decipher, from discernible trends, topics and subject matters that are saturating (at any given time) the collective consciousness of a diversified mix of people from across the globe.

Interestingly, this trait not only enables marketers to understand their target audiences better; it continually gives the human race an enhanced understanding of itself, which may at a certain point, consequentially enhance the quality of the choices that are made by the global society as a whole. (After, all knowledge is power, and self knowledge is infinite power - which contributes inordinately toward better decision making.)

Currently, my dominant passion is Artificial Intelligence. Although it might seem like an insurmountable goal, I envision bringing into being, at some point in the future, a trading and investment artificial intelligence that has the aggregate trading skills of the best traders in the world, and lacks the weaknesses of human beings.

Surprisingly, I was utterly clueless on how such a trading system could be designed, until I became acquainted with Twitter. I shan't go into the particulars, but from Twitter, one can collect volumes of time-series data from hundreds of thousands of individuals. This data is particularly useful in the realm of investments, and can be used for an infinite range of alpha-enhancing purposes, including the design of smarter investment algorithms.

Late at night yesterday, when I was extracting trends from the Twittersphere, I discovered that people Tweet about Money in a stable unvarying pattern (see Chart Below). I also discovered that people Laugh Out Loud (i.e. Lol) in their Tweets most between 6:30am - 7:30am. (see Chart below). Now, I need to figure-out the driving force behind these discerned trends, and I also need to find out how I can use the derived underlying insights to create robust quantitative trading algorithms.

Click On Image For Better Visibility

Enough said.