Stock market investors have always relied on traditional prediction strategies as the most accurate stock price forecasts. However, scientists now believe traditional strategies are not the most accurate! Twitter can predict stock prices better than any existing strategies, say University of California researchers. The researchers developed a model that leverages data from twitter to make predictions on stock prices. The model predicts the traded volume and value of the stock for the following day. The software finds data by scanning for tweets with keywords such as company names and other terms. Twitter last year announced its new [[xref:http://www.cio.com.au/article/400809/how_twitter_web_analytics_will_help_your_business/ |Web Analytics service|twitter applications]] that makes the microblogging service more attractive by quantifying results.
The software tabulates the amount of tweets and their relatedness and uses them to predict prices. The model performs better than all existing baseline strategies by between 1.4 percent and almost 11 percent. If adopted, the model could have serious positive implications for investors. It proved more effective than the Dow Jones Industrial Average during a four-month simulation. Many investors will be looking to sort out the immense data from social media and leverage it for profit. The researchers’ findings demonstrated that activity in Twitter has a direct correlation to stock prices and traded volume. Existing prediction research on stock prices through social media has always focused on sentiment, for instance the positive or negative effects of Tweets.
However, the University of California research goes beyond sentiment and analyzes volume of Tweets and the correlation between tweets, topics and users. This study also focused on individual stocks, as opposed to traditional researches that focus on overall stock market indices. The study was led by Professor Vagelis Hristidis, a specialist in data mining research that focuses on finding patterns in large data sets. The test run used daily closing price and the number of trades for 150 randomly selected companies in the S&P 500 Index for the first half of 2010. Separately, Chicago based Narrative Science recently developed a computer program capable of writing upnews stories by gathering data from tweets.