We deal with the optimal portfolio problem in discrete-time setting. Employing the discrete It\^o formula, which is developed by Fujita, we establish the discrete Hamilton-Jacobi-Bellman (d-HJB) equation for the value function. Simple examples of the d-HJB equation are also discussed.
We deal with numerical computation of the nonlinear partial differential equations (PDEs) of Black-Scholes type which incorporate the effect of transaction costs. Our proposed technique surmounts the difficulty of infinite domains and unbounded values of the solutions. Numerical implementation shows the validity of our scheme.