Concept
Finite difference methods for option pricing are numerical methods used in mathematical finance for the valuation of options. Finite difference methods were first applied to option pricing by Eduardo Schwartz in 1977. In general, finite difference methods are used to price options by approximating the (continuous-time) differential equation that describes how an option price evolves over time by a set of (discrete-time) difference equations. The discrete difference equations may then be solved iteratively to calculate a price for the option. The approach arises since the evolution of the option value can be modelled via a partial differential equation (PDE), as a function of (at least) time and price of underlying; see for example the Black–Scholes PDE. Once in this form, a finite difference model can be derived, and the valuation obtained. The approach can be used to solve derivative pricing problems that have, in general, the same level of complexity as those problems solved by tree approaches. As above, the PDE is expressed in a discretized form, using finite differences, and the evolution in the option price is then modelled using a lattice with corresponding dimensions: time runs from 0 to maturity; and price runs from 0 to a "high" value, such that the option is deeply in or out of the money. The option is then valued as follows: Maturity values are simply the difference between the exercise price of the option and the value of the underlying at each point (for a call, e.g., ). Values at the boundaries - i.e. at each earlier time where spot is at its highest or zero - are set based on moneyness or arbitrage bounds on option prices (for a call, for all t and as ). Values at other lattice points are calculated recursively (iteratively), starting at the time step preceding maturity and ending at time = 0. Here, using a technique such as Crank–Nicolson or the explicit method: the PDE is discretized per the technique chosen, such that the value at each lattice point is specified as a function of the value at later and adjacent points; see Stencil (numerical analysis); the value at each point is then found using the technique in question; working backwards in time from maturity, and inwards from the boundary prices.