Parallel Genetic Programming and application to financial forecasting

B. Chopard M. Oussaidene, M. Tomassini and O. Pictet

This research is supported by the Swiss National Science Foundation
Many problems in industry and economics require a very difficult optimization calculation. We would like to show that massively parallel computers, with the use of evolutionary algorithms such as genetic algorithms, provide a way to consider situations whose difficulty is much larger than what can be addressed in a reasonable amount of time with standard optimization techniques.

We have selected a problem which presents a considerable interest from the point of view of financial predictions and which, also, is representative of a class of very hard optimization problems. More precisely, this application concerns the construction of search algorithms for indicators of the future price movements of currency exchange rates by treating very large data sets of time series data.

We propose to address this particular problem in the framework of the parallel genetic programming and to elaborate a basic library of programs for solving efficiently other optimization problems such as those which are common in practical industrial and economical applications.

Some papers:

Other papers using the evolutionary approach:

GP Bibliography (from J. Koza and W.B. Langdon):