Quantcast

Twitter Analysis Provides Stock Predictions

April 4, 2011

Economists at the Technical University of Munich (Technische Universitaet Muenchen, TUM) have developed a website that predicts individual stock trends. To this end, economists are using automatic text analysis methods to evaluate thousands of daily Twitter microblog messages, so-called “tweets”. On www.TweetTrader.net, current forecasts are available for all S&P 500 listed stocks.

The share price of a stock reflects investor and analyst opinions about its prospects and indicates whether positive or negative developments are on the horizon. The micoblogging platform Twitter has become an important medium for the exchange of such viewpoints. Thousands of stock-related messages are broadcasted every day via Twitter. Twittering investors mark tweets according to company stock symbols, for example, “$AAPL” for the U.S. computer company Apple.

In a study, TUM economists showed that the sentiment from Twitter messages develops similar to the stock market and even leads by a day. The Munich-based economists analyzed 250,000 Twitter messages written in a six-month period and related to S&P 500 listed companies. The result: If an investor had oriented his share purchases according to the Twitter sentiment in the first half of 2010, he would have achieved an average rate of return of up to 15 percent.

The TUM economist Timm Sprenger explains, “If a Twitter user often gives good stock recommendations, he will, as a rule, have more followers and will be ‘retweeted’ (i.e., quoted) more often by other users. Hereby, tweets with good recommendations are affirmed and receive greater weight in the overall analysis.”

The study was the basis for the development of the website TweetTrader.net where the real-time sentiment for individual stocks can be accessed.

On the Net:

Stock Microblogging Forum TweetTrader: http://tweettrader.net/

The TweetTrader-Studie is available at the Social Science Research Network (SSRN): http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1702854




comments powered by Disqus