The Cox--Ross--Rubinstein model is a discrete-time frictionless market with a positive stock price and money-market account , where . In each step the stock is multiplied by or , with and both moves having positive probability. The usual model has independent moves, permits borrowing, lending and short sales, and ignores transaction costs and dividends. It gives a finite binomial tree for pricing claims by self-financing replication.[ref]
Option Pricing: A Simplified Approach by John C. Cox and others. Section3, manuscript pp.4–7, market assumptions and one-period replication
The market is arbitrage-free exactly when . Indeed the risk-neutral up probability is then , making the discounted stock a martingale. If , borrowing to buy the stock gives a nonnegative one-step gain and a positive gain in at least one possible state; if , the reversed position does so. Both possibilities contradict absence of arbitrage.
Discounting by the money-market account gives . Its up and down multipliers are and , so the same inequality becomes . Keeping these discounted multipliers separate from prevents a missing discount in a payoff valuation formula. The model is useful because two terminal states and two traded assets let one solve replication by two linear equations, without estimating a physical expected stock return.
A binomial price tree lists the possible stock prices in the Cox--Ross--Rubinstein model. Starting at , each step multiplies the price by or , with . After steps and up moves,Constant multiplicative steps make the tree recombine: an up move followed by a down move reaches the same price as a down move followed by an up move. Thus level has prices despite possible paths.
For , , , the second-level prices are . With risk-free multiplier , the no-arbitrage up probability is . A European call with strike 100 pays at that level, so its model value is . This probability is determined by replication and discounting, not a forecast of the physical chance of an up move.[ref]
Option Pricing: A Simplified Approach by John C. Cox and others. Section 3, binomial replication and multiperiod valuation
For a time step , choose the undiscounted stock multipliers and money-market multiplier asFor sufficiently small , and the risk-neutral up probability is . A one-step log-stock increment takes values ; expanding gives mean and variance . The Donsker invariance principle, in its triangular-array form, gives the risk-neutral log-price limitThus prices converge to the geometric Brownian motion used in the Black-Scholes model.[ref]
Option Pricing: A Simplified Approach by John C. Cox and others. Section5, manuscript pp.17 onward, small-step limits
The parent model's discounted multipliers are instead and . Their log-price drift loses the term, as required for the discounted price to be a martingale. Convergence of claim prices also needs the payoff functional to be continuous at almost every limiting path, together with Uniform integrability when payoffs are unbounded. A process limit alone does not justify taking expectations of every payoff.
The unique equivalent martingale measure in the one-stock Cox--Ross--Rubinstein model assigns each up move probabilityand each down move probability . Under this measure the discounted stock is a martingale.
Let be the undiscounted stock, the money-market account, and the risk-neutral measure of the Cox--Ross--Rubinstein model. A terminal payoff in currency units has arbitrage-free valueEquivalently, the discounted claim has discounted value . These are different unit conventions for the same valuation. For a European payoff and ,The outside discount and the undiscounted stock multipliers must appear together. If a payoff is instead expressed using the discounted stock, its function must be transformed accordingly.[ref]
Option Pricing: A Simplified Approach by John C. Cox and others. Section3, manuscript pp.7–11, risk-neutral discounted expectation and multiperiod formula
For , , , , a one-step European call option with strike100 pays20 or0. Its risk-neutral up probability is , so its value is . Holding one half-share and borrowing about38.0952 reproduces both payoffs. This replication explains why the price uses risk-neutral weights rather than the investor's physical up probability.
For a payoff with , the pair is a sufficient Markov state for valuation. If an independent restarted stock path has initial value , terminal value , and running maximum , then the continued payoff is The prices in this restart identity are undiscounted. Valuation still discounts the conditional payoff expectation by , as in the parent article.
Articles by others on the same topic
There are currently no matching articles.