The Fundamental Theorem of Asset Pricing Explained
This article provides a clear examination of the Fundamental Theorem of Asset Pricing (FTAP), a cornerstone of modern financial mathematics. It explains the core concepts of arbitrage and martingale measures, detailing how the theorem bridges the gap between economic intuition and probability theory to determine fair asset prices in financial markets.
What is the Fundamental Theorem of Asset Pricing?
The Fundamental Theorem of Asset Pricing provides the necessary and sufficient conditions for a financial market to be free of arbitrage. Developed through the work of economists and mathematicians like Stephen Ross, Michael Harrison, David Kreps, Freddy Delbaen, and Walter Schachermayer, the theorem is actually split into two parts:
- The First Fundamental Theorem: A market is free of arbitrage if and only if there exists at least one equivalent martingale measure (also known as a risk-neutral measure).
- The Second Fundamental Theorem: An arbitrage-free market is complete (meaning every contingent claim or derivative can be perfectly hedged) if and only if the equivalent martingale measure is unique.
To understand how the First Fundamental Theorem links arbitrage to martingale measures, it is necessary to define both concepts individually before examining their mathematical connection.
The Concept of Arbitrage
In finance, arbitrage is essentially a "free lunch." Mathematically, an arbitrage opportunity is a trading strategy that meets three criteria:
- It costs nothing to initiate today (zero initial capital).
- It has zero probability of losing money in the future.
- It has a strictly positive probability of making money in the future.
In a well-functioning, efficient market, arbitrage opportunities disappear almost instantly as traders exploit them, driving prices back into equilibrium. Therefore, pricing models assume the "No-Arbitrage" (NA) condition. In continuous-time mathematics, this is formalized as "No Free Lunch with Vanishing Risk" (NFLVR).
Martingales and Equivalent Measures
To price assets without arbitrage, mathematicians use the concept of a martingale from probability theory.
A martingale is a stochastic (random) process that models a totally fair game. In a martingale, the expected future value of a variable, given all past information, is equal to its exact present value. If a stock's price process is a martingale, your best guess for its price tomorrow is its price today; there is no predictable upward or downward drift.
In the real world, stock prices are not martingales. Investors demand a risk premium, meaning stocks generally have an upward drift over time. This real-world environment is governed by the physical probability measure (denoted as \(\mathbb{P}\)).
To use martingale mathematics, we must shift to a theoretical environment governed by a risk-neutral probability measure, or Equivalent Martingale Measure (denoted as \(\mathbb{Q}\)).
- Equivalent means that \(\mathbb{P}\) and \(\mathbb{Q}\) agree on what is possible and impossible (events with zero probability in the real world also have zero probability in the risk-neutral world).
- Martingale Measure means that under this new set of probabilities \(\mathbb{Q}\), the discounted price of every asset in the market behaves as a martingale.
By adjusting the probabilities of good and bad outcomes to absorb the risk premium, the expected return of all assets under \(\mathbb{Q}\) becomes equal to the risk-free interest rate.
How the Theorem Links No-Arbitrage to Martingale Measures
The brilliance of the First Fundamental Theorem of Asset Pricing lies in proving that the economic condition of "no arbitrage" is mathematically identical to the existence of an equivalent martingale measure. This link works in two directions:
1. From Martingale Measure to No-Arbitrage (The Easy Direction)
Suppose an equivalent martingale measure \(\mathbb{Q}\) exists. Under this measure, the discounted value of any self-financing trading portfolio is a martingale.
By the definition of a martingale, the expected future value of the portfolio under \(\mathbb{Q}\) must equal its initial value. Suppose you start with a portfolio worth $0 today. Because of the martingale property, its expected value in the future must also be $0.
For a random variable to have an expected value of $0, and assuming it can never be negative (no risk of loss), the variable must be $0 everywhere. It cannot have a positive probability of being greater than $0, because that would pull the average above $0. Since \(\mathbb{P}\) and \(\mathbb{Q}\) are equivalent, if there is zero probability of a positive payoff under \(\mathbb{Q}\), there is also zero probability of a positive payoff under the real-world measure \(\mathbb{P}\). Therefore, no arbitrage can exist.
2. From No-Arbitrage to Martingale Measure (The Hard Direction)
Proving that the absence of arbitrage guarantees the existence of a martingale measure requires functional analysis, specifically the Separating Hyperplane Theorem (or the Hahn-Banach theorem in infinite-dimensional spaces).
Imagine a geometric space of all possible future payoffs at a given time:
- Set A consists of all payoffs you can achieve starting with $0 today using valid trading strategies.
- Set B consists of all strictly positive payoffs (the definition of arbitrage).
If the market is free of arbitrage, you cannot achieve a strictly positive payoff starting with $0. Geometrically, this means Set A and Set B do not intersect.
Because these two sets do not overlap, the Separating Hyperplane Theorem states that we can draw a flat plane (a linear functional) strictly between them.
- Because this plane separates the positive payoffs from the zero-cost portfolios, the mathematical vector that defines the angle of this plane has strictly positive components.
- By normalizing these positive components so that they sum to 1, we create a valid set of probabilities.
This newly constructed set of probabilities is the Equivalent Martingale Measure \(\mathbb{Q}\). Because the zero-cost portfolios lie flat against this separating plane, their expected value under these new probabilities is zero, fulfilling the exact definition of a martingale.
Practical Application of the Link
The link established by the FTAP is the engine behind modern quantitative finance. Without it, pricing a financial derivative (like a stock option) would require knowing the real-world probabilities of stock movements and the subjective risk preferences of every investor in the market.
Because of the FTAP, quants can ignore real-world drift and investor psychology. As long as the market is free of arbitrage, they can switch to the risk-neutral measure \(\mathbb{Q}\), calculate the expected payoff of the derivative as if it were a simple martingale, and discount that value back to the present using the risk-free rate. The resulting price is guaranteed to be the only fair price that does not introduce arbitrage into the market.