8 Comments

Summary:

If Twitter can help you figure out which movie to see, can it also help predict which stocks will rise or fall? A German PhD student says his analysis shows it can, and he has set up a website to put his theories to the test.

4099481244_b0c62d93fb_z

Stock traders are constantly looking for ways to figure out what the markets are going to do so they can get an edge on everyone else — some even track seemingly unrelated factors like the weather to see which way the wind is blowing when it comes to the stock market. So it’s probably no surprise that now traders are trying to track the giant global conversation known as Twitter in order to figure out which stocks are becoming popular. But is there any evidence that this actually works? A German PhD student says that his analysis shows it can, and he has set up a website called TweetTrader.net to put his theories to the test.

The idea behind tracking Twitter is an appealing one: if millions of people are talking about the things they like or dislike via the service — whether those things are movies, books, food or celebrities — then it’s likely that some proportion of those tweets will be about stocks. Since “sentiment indicators” compiled by brokerage-firm research departments and economists are a big factor in the analysis done by many professional stock traders, the theory is that tracking the comments of individual investors and other observers in real time could provide similar kinds of information.

Timm Sprenger, a doctoral student in the school of management at the Technical University of Munich, recently published a research paper called “Tweets and Trades: The Information Content of Stock Microblogs,” in which he looked at the predictive value of tweets that mention stocks or the overall market. Sprenger and his research partner tracked and analyzed 250,000 tweets a day — over a six-month period — that fit that description, using semantic analysis to try and determine whether they were positive or negative (bullish or bearish, in stock-market terms).

According to Sprenger’s research, the sentiment rankings that his system extracted matched the ebb and flow of the Standard & Poor’s 500-stock index fairly closely, and appeared to predict movements in the market by more than a day. The researchers said that any investor who bought or sold using their analysis in the first half of 2010 would have achieved an average rate of return as high as 15 percent. At TweetTrader, which Sprenger set up recently with some fellow students, investors can track the sentiment rankings of various stocks and indexes based on the systems used in his research.

Sprenger’s paper is just the latest study to try and establish a connection between sentiment analysis of Twitter and the movements of the stock market. Last year, researchers from the University of Manchester and Indiana University published a report that looked at the predictive value of sentiment analysis extracted from Twitter (PDF link) compared to the movements of the Dow Jones Industrial Average. The study said that its system could predict the market index with 87-percent accuracy — and not long after the research was published, a hedge fund called Derwent Capital Markets announced that it was launching a fund that would trade based on these kinds of analysis.

The idea behind TweetTrader is similar to StockTwits.com, a service co-founded by entrepreneur Soren Macbeth and venture investor Howard Lindzon (see disclaimer below) as a way to track stock-related commentary and analysis. The site has progressed beyond just tracking or aggregating tweets with stock symbols, however, and has become what Lindzon calls a “social Bloomberg” or a “Facebook for finance” — a social network for those interested in stocks and trading. StockTwits also runs its own platform, something it decided to do after growing dissatisfied with Twitter’s repeated outages.

As I pointed out in my last post on Twitter-based stock analysis, the risk with Sprenger’s approach is that any open system such as Twitter can be “gamed” by those who want to promote certain stocks, just as message boards were in the early days of web-based investing — and just as Google’s search results are by SEO engineers. Anyone pursuing that kind of approach to investing is going to have to take that into account.

Disclosure: StockTwits is backed by True Ventures, a venture capital firm that is an investor in the parent company of this blog, Giga Omni Media. Om Malik, founder of Giga Omni Media, is also a venture partner at True.

Post and thumbnail photos courtesy of Flickr user Zach Supancic

  1. Very interesting. One can imagine other services that could arise based on the analysis of Twitter activity.

    Share
  2. Fascinating. I wonder if Twitter can also predict the rise and ecline of TV shows before (or more accurately than) Nielsen ratings are published. might help ad buyers choose the right platform and maybe eventually even at the right price?

    Share
    1. I know of a study that is looking at that for movies.

      Share
  3. ok ok…i will post on how bad the hedge fund idea is shortly…..its crafty marketing. If the returns are good, why would a hedge fund manager claim they came from twitter …if they are bad…case closed ….sentiment is lame. can be smart, can be dumb, but so far not imporessed. ideas are what matters

    Share
    1. Can’t wait for that

      Share
  4. Interesting…I’ve also heard of algorithmic trading machines mining Twitter for data. It’s definitely a useful stream, except it’s hard to say–especially with these algorithmic machines which compute millions of bits of data in seconds and make trades that cause fluctuations in the markets. You talked about people gaming Twitter, but I would think the algorithmic machines and their massive trades are increasingly the main factor in stock trading. See Flash Crash. However, the data in the story is convincing, and it’s intriguing nonetheless.

    Share
  5. I cannot see why not, the concept and Idea makes perfect sense. And I suppose its just that, common sense mixed with thorough research and experience that can make this so powerful? So does that mean that the man in the street will from now on affect how, what, when and where stocks are traded… interesting thought!

    Share
  6. I just wrote this about whethe Google can do the same? Maybe if you used both it would be more accurate as Google has a bigger sample size? http://searchinsights.wordpress.com/2011/04/08/can-search-engines-predict-the-future/

    Share
  7. Can Twitter Help You Predict the Stock Market? #twitter http://t.co/rURU5yd3

    Share
  8. Can Twitter Help You Predict the Stock Market? #twitter http://t.co/rURU5yd3

    Share
  9. Can Twitter Help You Predict the Stock Market? #twitter http://t.co/rURU5yd3

    Share
  10. Can Twitter Help You Predict the Stock Market? #twitter http://t.co/rURU5yd3

    Share

Comments have been disabled for this post