Warren: Financial Portfolio Management

In current practice, portfolio management is carried out by investment houses that employ teams of specialists for finding, filtering and evaluating relevant information. Based on their evaluation and on predictions of the economic future, the specialists make suggestions about buying or selling various financial instruments, such as stocks, bonds, mutual funds etc. Current practice as well as software engineering considerations motivate our multi-agent system architecture. A multi-agent system approach is natural for portfolio management because of the multiplicity of information sources and the different expertise that must be brought to bear to produce a good recommendation (e.g. a stock buy or sell decision).

The overall portfolio management task has several component tasks. These include eliciting (or learning) user profile information, collecting information on the user's initial portfolio position, and suggesting and monitoring a re-allocation to meet the user's current profile and investment goals. As time passes, assets in the portfolio will no longer meet the user's needs (and these needs may also be changing as well). Our initial system focuses on the ongoing portfolio monitoring process.

Click the following icon to view a diagram showing the main agents in the portfolio management task:

The portfolio manager agent is an interface agent that interacts graphically and textually with the user to acquire information about the user's profile and goals. The fundamental analysis agent is a task assistant that acquires and interprets information about a stock from the viewpoint of a stock's (fundamental) ``value''. Calculating fundamental value takes into consideration information such as a company's finances, forecasts of sales, earnings, expansion plans etc. The Technical Analysis agent uses numerical techniques such as moving averages, curve fitting, complex stochastic models, neural nets etc., to try to predict the near future in the stock market. The Breaking News agent tracks and filters news stories and decides if they are so important that the user needs to know about them immediately, in that the stock price may be immediately affected. The Analyst Tracking agent tries to gather intelligence about what human analysts are thinking about a company. These agents gather information through information requests to information agents. The information agents that we have currently implemented are the Stock Tracker agents that monitors stock reporting Internet sources, such as the Security APL, the News Tracking agents that track and filter Usenet relevant financial news articles (including CMU's Clarinet and Dow Jones news feeds), and the SEC (Securities and Exchange Commission) fillings of companies financial information tracker agent that monitors the EDGAR database. The information retrieved by these information agents is passed to the display agents which display in an integrated fashion the retrieved information to the user.

For more information, visit the Warren Project Page.



Multidisciplinary University Research Initiative (MURI)
Principal Investigator: Katia Sycara
Sponsored by: Office of Naval Research (ONR)
ONR Contact: Michael Shneier
© 1998 Carnegie Mellon Robotics Institute


Carnegie Mellon School of Computer Science Carnegie Mellon Robotics Institute DefenseLINK Office of Naval Research MURI Welcome University of Pittsburgh School of Information Sciences