Class-based distributed evolutionary algorithm for asset management and trading

Abstract

A server computer and a multitude of client computers form a network computing system that is scalable and adapted to continue to evaluate the performance characteristics of a number of genes generated using a software application. Each client computer continues to periodically receive data associated with the stored genes stored in its memory. Using this data, the client computers evaluate the performance characteristic of their genes by comparing a solution provided by the gene with the periodically received data associated with that gene. Accordingly, the performance characteristic of each gene may be updated and varied with each periodically received data. The performance characteristic of a gene defines its fitness. The genes may be virtual asset traders that recommend trading options. The genes may be assigned initially to different classes to improve convergence but may later be decided to merge with genes of other classes to improve diversity.

1 . A server computer system for use with a plurality of client computers each assigned to a respective class, comprising:
 a memory accessible to the server computer system and storing a server candidate pool having a plurality of individuals, each individual identifying a plurality of indicators and at least one corresponding action in dependence upon the indicators, each of the classes being associated with a respective subset of indicators in a plurality of indicators, not all of the subsets being the same;   a communications port through which the server computer system receives individuals from the client computers, including a first subset of the individuals all received from a first one of the classes; and   a processor configured to:
 determine whether the first class satisfies a predetermined convergence condition, and if so then to merge the first class with a second one of the classes to derive a merged class, at least some of the client computers that were assigned to the first class being re-assigned to the merged class, and 
 transmit to the client computers for further evaluation, each individual in the first subset of the received individuals; 
   wherein each individual in the first subset transmitted for further evaluation is transmitted for further evaluation to:
 a client computer in the first class, if the first class did not satisfy the predetermined convergence condition, and 
 a client computer in the merged class, if the first class did satisfy the predetermined convergence condition. 
   
     
     
         2 . The server computer system of  claim 1 , wherein the processor is further configured to:
 determine whether each of the received individuals is already stored in the server candidate pool; and   discard all of the received individuals having an updated fitness value which is below a predetermined server minimum fitness, wherein the updated fitness value for each received individual is,
 for individuals which are not already present in the server candidate pool, a fitness value received by the server computer system in conjunction with the received individual; and 
 for individuals which are already present in the server candidate pool, a combination of a fitness value received by the server computer system in conjunction with the received individual and a fitness value previously associated with the individual in the server candidate pool. 
   
     
     
         3 . The server computer system of  claim 2 , wherein the predetermined server minimum fitness is dependent upon a fitness of the least fit individual in the server candidate pool. 
     
     
         4 . The server computer system of  claim 2 , wherein the processor is further configured to store in the server candidate pool at least some of the accepted individuals. 
     
     
         5 . The server computer system of  claim 1 , wherein the individuals received by the server computer system through the communications port further include a second subset of individuals all received from the second class,
 wherein the second class is also determined by the server computer system to satisfy a predetermined convergence condition,   wherein the merging of the first class with the second class further includes re-assigning to the merged class at least some of the client computers that were assigned to the second class,   wherein the processor is further configured to transmit to the merged class of client computers for further evaluation, each individual in the second subset of the received individuals.   
     
     
         6 . The server computer of  claim 1 , wherein the merged class is associated with a union of the subset of indicators previously associated with the first class and the subset of indicators previously associated with the second class. 
     
     
         7 . The server computer system of  claim 1 , wherein one of the individuals transmitted for further evaluation is transmitted for further evaluation also to:
 an additional client computer in the first class, if the first class did not satisfy the predetermined convergence condition, and   an additional client computer in the merged class, if the first class did satisfy the predetermined convergence condition.