Previous  Top  Next

In Optimax, a generation is a collection of individuals all created at the same time and analysed as a set.

Generation - A group of individuals all born at about the same time. Also known as a population.

Each optimization is composed of several generations, usually 15 to 100, with each generation usually containing 50 to 200 individuals. Optimax begins the optimization using a randomly-generated set of individuals for the first generation, analyses them, and then creates a subsequent generation from them. It repeats this process throughout the optimization. You will specify how many individuals for Optimax to create for the first generation, and each subsequent generation will contain the same number of individuals as the first.

When you start the optimization, Optimax generates a set of completely random individuals which form the first generation. That means that the values for the genes of each individual in the first generation are completely random. Each of these individuals is passed to TradeStation one by one and evaluated, and a fitness value is sent back from TradeStation and stored with each individual. At the end of the first generation, Optimax generates a second generation by selecting individuals from the first generation for mating. The individuals are mated, a second generation produced, and then they are all sent one by one for processing to TradeStation and fitness values returned and stored. The process repeats until convergence is reached.

During optimization, every individual within a generation is guaranteed by Optimax to be unique. When each individual is created, no matter which method is used, it is compared with all other individuals of the generation in a "clone check" operation. If it is identical to another individual, Optimax uses one or more of three changeling methods: mutation, rival, or random, to create a unique individual.

When Mutation is chosen, a single gene is mutated at random and this process is repeated until the individual is unique within the population.

When Rival is chosen, roughly one-half of the genes of the individual are shifted up or down by one step to create a near likeness, or a rival to the existing individual. Each gene has a 50% chance of being shifted, and there is a 50% chance of an upward or downward shift. If, after shifting all genes and creating a rival, it is found that there is a clone in the population for this new individual, then the process is repeated until a clone is not found.

When Random is chosen, an entirely new, entirely random individual is generated.