Markov Processes and Their Role in Option Pricing
Summary
The answer defines the Markov property as conditional independence of future states from the process’s earlier history, given its current state. In an asset-price example, the next move depends on the current state rather than on the particular path that led there. This memoryless assumption can simplify modeling and option-pricing calculations.
The response distinguishes this property from the martingale condition used in arbitrage-free pricing: being Markovian alone does not make a process a martingale. It offers an intuitive rationale for modeling asset prices as Markovian but does not derive the assumption from market principles or establish that real asset prices always satisfy it. The exchange is a conceptual introduction, not a justification for choosing a particular stochastic model or a complete account of risk-neutral pricing.
Key ideas
- A Markov process makes the future conditionally dependent on the current state rather than the full history.
- The Markov property is also described as memorylessness.
- Markovian behavior can simplify financial modeling and option-pricing calculations.
- The Markov property alone does not imply that a process is a martingale.
- The answer presents an intuition for the assumption but does not prove that real asset prices are Markovian.
Tags
Full text
# Use of markov process in option pricing
# Use of markov process in option pricing
In several books on asset pricing and more particularly when it concerns option pricing, I see the use of Markov process, they argue the computation is made easier with such process. Is this sufficient to use Markov Model ?
For example, when we talk about martingale property, this is the result of a fundamental theorem in asset pricing and the proof shows us why it is true. For Markov process I did not see that and I wonder what are the foundations of this Markovian approach
Thank you a lot
## Answer by Jan Stuller (score 3, accepted)
https://quant.stackexchange.com/a/71077
The Markov property dictates that the future states of a stochastic process only depend on its current state, not any previous states. In a discrete setting, this can be written as:
$$\mathbb{P}(X_{n+1}=x|X_n=x_n,X_{n-1}=x_{n-1)},...,X_0=x_0)=\mathbb{P}(X_{n+1}=x|X_n=x_n)$$
In finance, we want asset prices like stocks to satisfy the Markov property: say the stock price was 100 last week and it's gone up by 10% to 110: the probability that it goes up by another 10% or that it goes down by 10% should not depend on what happened last week; it should only depend on the current state of the world.
The Markov property is also called "memorylessness": somewhat intuitively, asset's behaviour should not depend on the "path" it took to reach the current state; assets should not "remember" their past in order to drive their future states (if we could determine stock price based on the past, it would be easy to make money, wouldn't it? The fact that everyone is trying and most people fail over the long term suggests that indeed, real-world assets are Markovian).
The martingale property is important for derivative pricing, and is related to the cost of borrowing and lending money. Simply put, in "expectation" (where "expectation" represents the mathematical operator that sets the future arbitrage-free price), we want asset prices to equal to their current value compounded at the rate of borrowing money; indeed, as proven in the Fundamental Theorem of Asset Pricing.
PS: note that not every Markov process is a martingale, as discussed in this question.Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.