On Monday I had the opportunity to attend a conference at Goldman Sachs’ Dallas office. Among the dozens of money managers and investors who attended, a combined $1 trillion in assets was represented. The speakers were numerous, from famed economist Jan Hatzius, Goldman’s head of global economics, to Jeff Currie, global head of commodities research. Everyone was exceedingly smart and articulate, and I left the conference feeling recharged with much to think about.
One of the most fascinating takeaways was Goldman’s increased use of sentiment analysis tools. Basically what this means is sophisticated software trawls the internet in real time for public attitudes and opinions on companies, products, sectors, industries, countries—you name it. Sources can include press releases, news stories, earnings calls, blogs, social media and more. All of this data is gathered and analyzed, giving quants and other highly sophisticated investors a better idea of where tomorrow’s opportunities lie.
We have experience gauging sentiment using platforms designed by Meltwater and ScribbleLive, and I was pleased to see our efforts validated.
Goldman’s preferred system is Stanford’s CoreNLP, which is able to break down and analyze sentences in a number of different ways (and different languages to boot). Below is just a sampling of what the process looks like.