Special Issue of Computers & Operations Research Applications of OR in Finance

Recent years have seen the development of numerous applications for operations research models and algorithms in the financial world, as computer capacity and power were growing exponentially. Financial institutions, large corporations and research centres are increasingly devoting important resources to research and development in financial modelling and optimization. This special issue presents specific financial applications using operations research models and algorithms, covering the broad areas of investment decision support, asset pricing and estimation. A first paper, by D. Bertsimas and D. Pachamanova, suggests robust linear programming formulations for the multiperiod portfolio optimization problem. Their formulation allows for the inclusion of specific portfolio constraints, and particularly transaction costs, while keeping the mutiperiod problem computationally manageable. They report on experiments using simulated market data, comparing the risk/return performances of various formulations. The second paper, by M.J. Best and J. Hlouskova, also considers transaction costs, but in the context of a single period mean-variance model formulated as a quadratic programming problem. Accounting for transaction costs increases the dimension of the QP, and the authors propose an algorithm to efficiently solve this specific problem. They report on experiments comparing the computational efficiency of various algorithms using generated data. In the third paper, by L.Yu, S. Wang and K.K. Lai, the portfolio selection problem is formulated as a mean-varianceskewness trade-off model. The authors propose a neural network based method for simultaneously forecasting market behaviour and selecting trading strategies. Their approach is tested on historical market data, and the performance of forecasting models and trading strategies are compared in an out-of-sample experiment. The next paper, by R. Josa-Fombellida and J.P. Rincón-Zapatero, studies the optimal management of a pension fund using a stochastic optimal control model. The authors assume that the contributors to the fund have different salaries and that all elements of the funds are stochastic. They derive the properties of optimal allocation strategies and optimal fund contributions. They characterize the solution when salaries follow a geometric Brownian motion. The fifth paper, by J. de Frutos, presents a numerical method for the valuation of bonds containing embedded options. The pricing problem is set as the solution of a system of partial differential equations, which are approximated by a spectral method using Laguerre polynomials. Numerical experiments, using a CIR model for the short-term interest rate, show the method to be very efficient with respect to other existing numerical approaches. The next paper, by X. Li and Z. Wu, proposes an approximation method for the valuation of high-dimensional basket options. The assets are assumed to have mean-reverting prices. The approach proposed by the authors consists in approximating the basket by a single asset following a log-normal process, which can then be easily priced. Numerical experiments show this approach to be computationally efficient and accurate with respect to Monte-Carlo simulation and other approximation approaches. The seventh paper, by E. Errais and J. Sadowsky, is devoted to the pricing of a real option, namely the valuation of an investment opportunity preceded by a sequence of pilot phases giving information about the technical risks associated with the project. The authors analyse the value of the investment opportunity with respect to market volatility, learning speed and other determinants, and present an approximate dynamic programming algorithm to solve the pricing problem. An implementation of the algorithm on a simplified version of the model is presented and discussed.