Optimization of Shares Portfolio Using Genetics Algorithms and Value at Risk

The choice of an optimal portfolio of shares has long been a topic of major interest in finance . The basic idea is to maximize returns for a given risk or to minimize the risk for a given return. This choice is based on an arbitration risk / return. In this paper, an algorithm called dynamic MinVaRMaxVaL is proposed in order to select an optimal portfolio of stocks using value at risk (VaR) and genetic algorithms that represent stochastic optimization methods based on stochastic search techniques built on mechanisms of natural selection and natural genetics. The objective of our algorithm is to minimize risk and maximize portfolio value at the same time through two stages. The first step is to minimize the risk measured by the value at risk (VaR) for a given value of portfolio. While the second step, dynamically we maximizes the value of portfolio such as the result obtained is greater than the portfolio value set at the first stage and the risk resulting from the second stage is lower than that obtained from the first.