Technology has utterly reworked the "plumbing" of markets, dissemination of prices, orders, and clearing. A new frontier in financial technology is dealing with information used to make decisions that arrives as text or web content, as eInformation
More on eInformation (a slight update of an essay from 2002, by Dave Leinweber and Peter Tufano)
A Transformation in Financial Information
Information is the raw material for any investment or trading strategy, and technology can radically alter the information landscape. More than a century ago, a new technology - Transatlantic Telegraph Cables - had a major impact on the functioning of financial markets by suddenly bringing the prices of securities between New York and London in line with one another. The Internet may offer a similar revolution in communication and may influence financial markets. Recent years have witnessed the creation of new means by which information, opinions and analyses can be shared among investors. Information frictions are dramatically reduced. Information is impounded in prices much faster. Press releases used to be read by editors, now they are read by everyone. Local newspapers were read locally, now everything is global. Information sharing used to occur within small groups, now there are hundreds of millions of people reading and writing messages on the “Internet wall”. Since the late 90s, we have seen an astonishing growth of freely accessible information on the Net, an equally astonishing growth in the number of people able to act on this information. There is more and more information (and disinformation) available, and with over 20 million online brokerage accounts, it is impounded in prices at ever increasing speeds. This transformation is dramatic.
The e-Information Project
Hundreds of articles on the informational efficiency of financial markets have been published over the past quarter century in major finance and accounting journals. However the changes brought about by the net have created new opportunities to learn from empirical examination of the information in markets and the relationship of this information to the price formation process, to returns, to volatility, volume, and liquidity. Which types of information events lead market response, which types are responses to market activity?
The e-Information group brings together computer scientists and financial economists with the goal of reaching an understanding of the new media for information transmittal and the relationships between market and information. In particular, Internet-based stock message boards provide a real time window into the minds of some individual investors. Our research team, among others around the country, is seeking to understand how this new technologically-enabled form of communication works. Our research seeks to understand the salience of what we call "e-information": investor sentiment and disagreement, extracted from the Web using statistical language processing.
Below you will find some of our preliminary work, as well links to related commercial and academic work. We are currently working on the construction of a large dataset to test a variety of intriguing propositions about e-information and the stock market. If you find other related links, let us know and we can post them for other researchers to share.
- "e-information". Sanjiv Das, Asis Martinez-Jerez, and Peter Tufano. In this clinical study, we create a new micro-database of much of the information flow about four stocks for a period of six months, including a novel database of e-Information. We define e-Information as estimates of the intensity and dispersion of investor sentiment extracted from four major stock chat message boards. We combine this e-Information with other components of the traditional information set (formal press releases by the firms, filings, analyst revisions, and news stories available electronically) to create an intensive database of as close to the full information flow as is practical. We document certain patterns in the release of information and show that there are suggestions of links between our measure of e-Information and contemporaneous stock returns and implied volatility. While e-Information does not seem to predict subsequent stock price returns, it may be related to the implied volatilities taken from options prices. (Revision dated November 2001)
- "Yahoo! for Amazon: Opinion Extraction From Small Talk on the Web" Sanjiv Das and Mike Chen. This paper describes the algorythm used to extract sentiment from the web. In addition to the paper, additional resources are given below.
- Seminar slides (pdf)
- Sentiment Index Generator
- Example plots of stock prices and sentiment (n.b., stock price is measured in "dollars" and the sentiment is measured in a new metric, called "sents")
- "Three hundred years of stock market manipulation," David Leinweber and Ananth Madhaven, The Journal of Investing 10 (7/Summer 2001), 7-16. This paper shows how stock chat boards can be used to manipulate stock prices.
Commercial sites seeking to understand how e-information works:
Blogs dealing with eInformation:
Selected related research:
- Antweiler, W. and Murray Frank, 2001, Is all that talk just noise? The information content of internet stock message boards, unpublished manuscript
- "Catching the Buzz," Scientific American, November 2001, 30-32
- M. Hirshey, V. Richardson, & S. Scholz, "How Foolish are Internet Investors?" Financial Analysts Journal, Jan/Feb 2000, 62-69
- Tumarkin, Robert and Robert Whitelaw, 2001, "News or Noise? Internet message board activity and stock prices," Financial Analysts Journal 57, 41-51
- Wysocki, Peter, 1999, Cheap talk on the Web: The determinants of postings on internet stock message boards, unpublished manuscript
- The AEnalyst Project, by Victor Lavrenko, Matt Schmill, Dawn Lawrie and Paul Ogilvie, UMass Computer Science Department. This group is extracting sentiment from news stories.