Google Play icon

Predictions of social investing platforms are often incorrect, study shows

Posted June 6, 2014

Stock market consulting was exclusively controlled by large companies since the 19th century. However, the situation started to change and decentralized online communities are increasingly visible in recent years. However, the quality of their advices and their actual impact remains unclear.

NASDAQ stock market displays at Time Square. Author: bfishadow on Flickr. Year: 2007.

NASDAQ stock market displays at Time Square. Author: bfishadow on Flickr (CC BY 2.0). Year: 2007.

Scientists at the University of California, Santa Barbara and the Tsinghua University analyzed two influential online platforms: SeekingAlpha and StockTwits.

Two websites have profound differences. “SeekingAlpha content is mainly contributed by roughly 8000 registered contributors, and articles are vetted by an editorial board before publication on the site,” the researchers say. StockTwits is a twitter-like page. Anyone can write short messages and contents of the posts are not monitored.

“We start by studying how well each article predicts the future price trends of the stocks it discusses,” the scholars say. Data analysis revealed that most of the contributors perform pretty poorly. Their predictions are only a little bit more reliable than a random guess. However, more detailed investigation revealed that some of the authors writing for the SeekingAlpha offer highly reliable forecasts.

“More importantly, these authors can be identified not only by their statistical performance, but more easily by the feedback their articles generate from other users,” the researchers add. In fact, their prognoses outperform most of the predictions, which are supplied by SeekingAlpha competitors.

In addition, Gang Wang and his colleagues discovered that users are able to identify entries which are manipulative and signalize them to others users via negative comments. “While only a small portion of users can deliver valuable content, others in the crowd help by indirectly identifying the valuable content through their interactions,” the scientists conclude.

Original research article: arXiv:1406.1137

Featured news from related categories:

Technology Org App
Google Play icon
87,029 science & technology articles

Most Popular Articles

  1. You Might Not Need a Hybrid Car If This Invention Works (January 11, 2020)
  2. Toyota Raize a new cool compact SUV that we will not see in this part of the world (November 24, 2019)
  3. An 18 carat gold nugget made of plastic (January 13, 2020)
  4. Human body temperature has decreased in United States, study finds (January 10, 2020)
  5. Donkeys actually prefer living in hot climate zones (January 6, 2020)

Follow us

Facebook   Twitter   Pinterest   Tumblr   RSS   Newsletter via Email