ABSTRACT
18Of late, the field of portfolio optimization (the process of selecting the proportions of diverse assets existing within a portfolio and building the portfolio to be the best in relation to some criterion) in the modern capital market has assumed paramount importance as a field in business intelligence thanks to the evolution of the multiobjective optimization of the market risk–return paradigm. The downside risk can be ascertained with a very common method within a portfolio known as value at risk (VaR), which, in turn, can be explicated as the pth percentile of return of a defined portfolio during the termination of the planning skyline. The conditional value at risk (CVaR) is a more robust expedient for determining the defined unit of risk of a portfolio in volatile market conditions. The soft computing paradigm is efficient in handling real-life uncertainties. It entails several tools and techniques, namely, neural networks, the concept of fuzzy logic, and evolutionary computation measures. Applications of three different soft computing-based metaheuristic approaches to risk minimization leading to portfolio optimization using particle swarm optimization (PSO), ant colony optimization (ACO), and differential evolution (DE) techniques centering on optimizing the CVaR measure under different market conditions based on several objectives and constraints are reported in this chapter. The proposed approaches are proven to be reliable on a collection of several financial instruments as compared to their VaR counterparts. The results obtained show encouraging avenues in determining optimal portfolio returns.
