My Techlab Webpage Charles Vu 2002-2003 Project Proposal

Title: An Analysis of Option Pricing Models

Problem Statement: The stock options market is a trillion dollar industry and many models exist which attempt to price stock derivatives The Nobel-prize winning Black-Scholes model is the benchmark of option pricing models. However, wheter or not the Black-Scholes model accurately predicts the intrinsic value of an option is still up to debate. Other widely accepted models exist, including the Cox, Ross & Rubenstein binomial model, Black's apporximation, and the Barone-Adesi and Whaley quadratic approximation.

Purpose: The goal of this project is to determine which option pricing model most accurately calculates option prices. If it can be determined which procing model is best, a deviation between an actual option price and the theoretical could lead to a greater than 50/50 odds of a successful trade. The models will be tested numnerically; the theory and math behind each model will not be discussed.

Scope of study: Only the most popular pricing models will be studied. Extensive back-testing will be the primary method of testing; past stock and optino data as well as some process of transferring this data will be needed. A limitation is that all pricing models assume random market movement.

Background and review of current literature/research in this area: Many research papers have been written discussing the accuracy of the theory bethind each of the most popular pricing models. However, there has not seem to have been any numerical analysis of the models; to see how well they performed in real-life markets.

Procedure and Methodology: The primary programming language will be C++. Initially, the equations of the various models will need to be programmed. Past stock and option data will then be collected for testing. Simple call and put strategies will be employed. The "best" model will be the one that yields a portfolio with the closest to no return. There is a possibility that the market price is the best model, and will also be tested.

Expected Results & Value to Others: The "best" pricing model should be identified following the project. Graphs of the testing will be presented. Once a "best" pricing model is identified, other researchers can hypothesize as to why it is most accurate.

Back to home