Pablo Almaleck, Davide Anghinolfi,
Nicolo` Bianco, Claudio Castellini
LIDOlab, DIST, University of Genova (Italy)
ailab, DIST, University of Genova (Italy)


Stock trading and genetic algorithms


We will be talking, in a quite informal way, of a recent application of Genetic Algorithms (GA) to Stock Trading. In particular, we will focus upon: (1) building agents which are trained on historical pricing data and are then forced to continually evolve to follow the trends of a single stock; (2) building portoflio managing agents which are trained, once and for all, upon a short historical time window and then select appropriate parameter values from an admissible range. We will be illustrating our ideas with live demos and graphs; preliminary results show that a careful choice of the trading strategies can produce good results, i.e., making the winning agents much richer than the bank, the buy-and-hold strategy or the trend-following agent.