TheStockSonar.com: Track Stocks Based On Media Sentiment

Next Story

The Case for Bartz NOT Getting Fired from Yahoo

We all have had the feeling that somehow stock prices are related to good and bad press. A story goes out in BusinessWeek (or TechCrunch!) and a few days later the stock goes up or down. But until now it was only a gut feeling, something you wouldn’t want to bank your nest egg on. Digital Trowel’s TheStockSonar.com is a tool that essentially brings some statistical analysis to bear on that problem and the results are – if not mesmerizing – quite interesting.

The site is basically a stock look-up system. You type in a stock, say APPL, and you get a graph of good and bad sentiment as well as the stock price. The sentiment analysis is all automatic and is based on reports from major news organizations. If the sentiment is good (“We believe this stock is going up”), the green line goes up. If the sentiment is bad (“BP kills baby seals”), the red line goes up. However, TheStockSonar has special semantic technology that can recognize false positives (“BP is great at killing baby seals”). In one demo they showed how the software found real negative sentiment in a paragraph that would normally be seen as positive by a machine (“The shares fell 13%, despite a record quarter sales growth and better-than-expected earnings.”).

While I wouldn’t recommend investing your millions based on this product just yet, Digital Trowel hopes to add this as a decision-making tool for traders and they have already created a number of implementations that stream a live feed of stock sentiment to automated trading systems, thereby allowing robots to understand general sentiment.

The system could also work for other types of articles including sports betting, media, and gossip. The company is based outside of Tel Aviv and is funded primarily by a single large investor. The team consists of a bunch of semantics nerds and Prof. Ronen Feldman, one of the giants in text recognition and learning.

The company is also working on creating automated corporate biographies based on publicly available information. This would allow CRM systems aimed at C-level execs to automatically update when that person switches jobs or changes positions.
The service is free for now although they are in talks with many clients to offer their feed and sentiment analysis on a more formal basis.

blog comments powered by Disqus