A few Prediction Company references

This web page collects a few Prediction Company references available on the World Wide Web.

• The Business of Model Based Trading (PDF format), Prediction Company, April 2001, Published on the Quantitative Work Alliance for Applied Finance, Eduction and Wisdom web site (www.qwafafew.org)

  This was from a talk titled From Roulette to Finance given in Boston by Norman Packard, the CEO of Prediction Company, on April 10, 2001.

• Black Box: What Happens when maverick physicists in New Mexico set out to predict the markets? (PDF format)
  by Thomas Bass, The New Yorker, April 26, 1999
  This article is based on the book The Predictors and appeared before publication of the book. This article will give the reader a feel for Bass' style of journalism and the general content (or lack there of) in The Predictors.

• Prediction Machinery, Chapter 22 of Out of Control by Kevin Kelly, Perseus Publishing, May 1995

  Kevin Kelly was one of the founders of Wired Magazine. Out of Control was written at the dawn of the Web age, when the seeds of the dot-com bubble were germinating. This chapter discusses the early days of Prediction Company (via interviews with company founders Doyne Farmer and Norman Packard), along with prediction in general. Kelly is an engaging popular science writer.

• Towards a Theory of Autonomous, Optimising Agents by Vince Darley, Phd Thesis, Harvard University, June 1999

  Financial time series seem to be both non-linear and chaotic, so there has been a significan amount of interest in using techniques from fractal and chaos theory to predict financial time series. Although a great deal has been written in this area, no successes have been reported.

  2.5.2 Deterministic Chaos

  There have been many attempts to detect the presence of deterministic chaos in complex real-world phenomena. In many engineering cases such studies have proved very useful. Curiously all attempts to approach the stockmarket from this perspective have either failed, or been largely inconclusive. Given the hypothesis that deterministic chaos exists in the timeseries observed in various high-volume trading markets, the idea is simple: search for and isolate small 'pockets of predictability' within the turbulent whole. Within those pockets, accurate predictions may be made for a short period of time, and if suficient money lies behind those predictions, much more can be made. With this scheme it is only necessary to make trades within those pockets, so it is only required that they can be identified quickly, which in turn requires that they be reasonably low-dimensional. As an example, the 'Prediction Company' founded by D. Farmer and N. Packard, sought to make use of that approach. They discovered that it was not possible with the best techniques they could design, and concluded that deterministic chaos could not be usefully harnessed in the markets, although it may well exist.

• Toward Agent-Based Models for Investment (PDF format) by J. Doyne Farmer, 2001, Benchmarks and Attribution Analysis, Association for Investment Management and Research.

  Doyne Farmer was one of the founders of Prediction Company. He is now at the Santa Fe Institute.

  Agent based modeling has been suggested as a non-linear modeling technique. Agents have memory and can embody complex algorithms, including learning algorithms. Agents can model complex systems, like markets. Although these systems can show predictive power, the complexity of agent based systems can make the behavior of these models difficult or impossible to understand. When building financial models, that are used for real trading, this can be a critical issue. If the predictions from an agent based model have large gains or losses, it can be difficult or impossible to understand the factors in the model that lead to these results. Nor is it easy or perhaps even possible to establish an expected range of profit and loss.

• The Phynancier by Thomas A. Bass, Wired Magazine, January 1997

  This is an article on David E. Shaw, the founder and principle of D.E. Shaw and Co., an investment and technology development company. D.E. Shaw as a pioneer in computer driven quantitative finance and, presumably, continues to use these techniques today. At least conceptually D.E. Shaw is the fore runner of Prediction Company.

  After this Bass article was published, the D.E. Shaw investment fund experience a large loss during the 1998 Asian market meltdown. Bank America was a large D.E. Shaw investor. They lost $372 million (US). Along with the demise of Long Term Capital Management, this provides a useful cautionary tale.

  D.E. Shaw is also famous for being the last employer of Jeff Bezos, before he founded Amazon.com. D.E. Shaw founded Juno (juno.com) as well.

  A few additional D.E. Shaw references:

  • Inside D.E. Shaw, Derivitives Strategy, 1898

    This "before the fall" article recounts D.E. Shaw's newly established relationship with Bank of America.

  • Profiting from the appliance of science by John Authers, Financial Times, Mar 26, 2001 (Microsoft .doc format)

  • Judge Approves Deal For $490 Million In Bank Of America Suit; Attorneys Want $122 Million In Fees by Peter Shinkle, St. Louis Post-Dispatch. October 2, 2002

    This article discusses the Bank America settlement of a class action law suit involving the D. E. Shaw losses (and the outrageous legals fees).