Genetic algorithms are problem-solving methods that mimic natural evolution processes.
Traders use them to predict stock prices and identify the best values for security parameters. They can also find the best combined parameter values in a trading rule, and traders build them into artificial neural network models to pick stocks. Artificial neural networks function like brain neurons and can adapt to changing criteria.
Genetic algorithms come from vectors, which are lines on a graph with direction and magnitude. Each trading rule’s parameters are represented with a one-dimensional vector. Each vector is like a chromosome, and each parameter is like a gene, which is modified using natural selection.
Three genetic operation types can be performed: crossovers, mutations and selections.
Crossovers are seen in biology, like when a child takes on her mother’s characteristics.
Mutations are small, random changes that maintain genetic diversity from one generation to the next.
Selections occur when individual genomes are chosen for later breeding.
These operators are part of a five-step process.
First, parameters are established with a certain number of elements each. Next, the parameters that increase desirable results are selected. Then mutation and crossover operators are applied to the selected parameters to generate offspring. After that, the offspring are recombined with the current population to form a new population. Finally, all steps are repeated, resulting in better trading parameters over time.
Read more: What are Genetic Algorithms? - Video | Investopedia http://www.investopedia.com/video/play/what-are-genetic-algorithms/#ixzz3tmmsuDIn
Follow us: Investopedia on Facebook
