Using genetic algorithms to search for an optimal investment strategy

POSTER

Abstract

In this experiment we used genetic algorithms to search for an investment strategy by dividing capital among different stocks with varying returns. The algorithm involves having a ``manager'' who divides his capital among various ``experts'' each of whom has a simple investment strategy. The expert strategies act like genes, experiencing mutation and crossover, in a selection process using previous returns as the fitness function. When algorithm was run with test data where the optimal strategy favored non-uniform investment in one stock it consistently beat a simple buy hold. However when the algorithm was run on actual stock data the system overwhelmingly stabilized at a population that closely resembled a simple buy hold portfolio, that is, evenly distribute the capital among all stocks.

Authors

  • Edward Mandere

    Bowling Green State University

  • William H. Rauckhorst

    Gothenburg University, Denison University, Bowling Green State University, NASA Glenn Research Center, Ohio Wesleyan University, Nitronex Corporation, 2305 Presidential Drive, Durham, NC, AFRL/MLPS, Wright-Patterson AFB, OH, Miami University, Ursinus College, Purdue University Calumet, GSI, Florida State University, Carnegie Mellon University, Dept. of Health Sciences, Cleveland State University, Physics Dept., Cleveland State University, University of Rochester, OMI College of Applied Science, University of Cincinnati, Davis and Elkins College, WV, Schlumberger-Doll, Korea Institute of Science and Technology, University of Cincinnati, Lock Haven University of Pennsylvania, Oak Ridge National Laboratory, Oak Ridge, TN 37831-6372, Albion College, Albion, MI 49224, University of Toledo, Toledo, OH 43606, Youngstown State University, Joint Institute for Nuclear Research (Dubna), Naval Research Laboratory, ManTech SRS Technologies, Inc., Ecopulse, Inc., Army Research Laboratory, University of Surrey, The Ohio State University, Univ. Akron, Owens Community College, Toledo, OH