How Jump Diffusion Model works?

In jump diffusion model, prices are permitted to jump discontinuously in addition to the continuous Brownian motion (the diffusion) observed in Black-Scholes models. Typically, the jump’s timing is random, and a Poisson process represents this. Furthermore, the jump’s size may be at random. The values of calls and puts rise as the jump frequency increases (all other parameters staying the same). Binary and other option prices are subject to fluctuations.

Jump Diffusion Model

A Short Example

A stock moves in a random walk that is lognormal. Each month, you take a dice roll. Should a one be rolled, the stock price will rise sharply. The amount of this jump is determined by drawing a random number out of a hat. (This is not a very good illustration because the Poisson process is not a monthly occurrence; rather, it is a continuous process.)

A Deep Dive into Jump Diffusion Model and Option Values

The Poisson process is represented by the symbol dq, which represents the jump in the random variable q from time t to time t + dt. With probability 1 - \lambda dt, dq=0; with probability \lambda dt, it is 1.
Observe how the likelihood of a jump increases with the potential jump duration, dt. The intensity of the process is represented by the scale factor \lambda; the bigger \lambda, the more frequent the jumps.

An interest rate, volatility, or equity price are examples of discontinuous financial random variables that can be modelled using this method. While studies on pure jump processes as financial models have been published, it is more common to mix jumps with traditional Brownian motion. For example, the equity model is commonly believed to represent

dS = \mu S dt + \sigma S dX + (J - 1)S dq

As previously mentioned, dq has an intensity of \lambda, and the jump size, J-1, is typically assumed to be random. Jump-diffusion models are useful for capturing the fat tails observed in return data, as well as for simulating the real-world occurrence of variable discontinuities.

A model for option prices is derived from the model for the underlying asset. Usually, this model will take the form of an integro-differential equation, where the integral term denotes the probability of the stock leaping discontinuously across a finite distance. Sadly, markets with these kinds of surges are incomplete, thus it is impossible to hedge options to remove risk. Therefore, one needs either introduce more securities to hedge or make certain assumptions about risk preferences in order to develop option-pricing equations.

History of Jump Diffusion Model

The jump diffusion model was first proposed by Robert Merton. He arrived at the following formula for the values of equity options.

\dfrac {\partial V}{\partial t} + \dfrac {1}{2}\sigma ^{2}S^{2}\dfrac{\partial ^{2}V}{\partial S^{2}}+rS \dfrac{\partial V}{\partial S} - rV + \lambda E[V(JS,t)-V(S,t)] - \lambda \dfrac{\partial V}{\partial S}SE[J-1]=0

The expectation taken across the leap size is denoted by E[ยท]. This equation expresses the expected value of the discounted payout in probability terms. The true measure for the leaps is higher than the risk-neutral value for the diffusion.

In the unique situation where the logarithm of J has a normal distribution, there is an easy way to solve this equation. In the event where J‘s logarithm has a normal distribution with a standard deviation of \sigma and we write

k = E[J-1]

Consequently, the cost of a non-path-dependent European option can be expressed as

\sum _{n=0}^{\infty}\dfrac{1}{n!}e^{-{\lambda}'(T-t)}({\lambda}'(T-t))^{n}V_{BS}(S,t;\sigma _n, r_n)

In the above equation,

{\lambda}' = \lambda (1+k), \ \sigma _n^{2} = \sigma ^{2}+\dfrac{n{\sigma}'^{2}}{T-t}

and,

r_n = r - \lambda k + \dfrac {n\ ln(1+k)}{T-t}

and V_{BS} stands for the option value in the absence of leaps in the Black-Scholes algorithm. This formula can be understood as the total of discrete Black-Scholes values, each of which is weighted based on the likelihood that n leaps will have occurred before to expiry and assumes that there have been n jumps.

While other models, like stochastic volatility, struggle to capture steepness in volatility skews and smiles for short-dated options, jump diffusion model is able to do so with relative ease.

Related Readings

Leave a Reply

Your email address will not be published. Required fields are marked *