Python for Finance: Analyze Big Financial Data

(Elle) #1

Fundamental Theorem of Asset Pricing


The Fundamental Theorem of Asset Pricing is one of the cornerstones and success stories


of modern financial theory and mathematics.


[ 61 ]

The central notion underlying the


Fundamental Theorem of Asset Pricing is the concept of a martingale measure; i.e., a


probability measure that removes the drift from a discounted risk factor (stochastic


process). In other words, under a martingale measure, all risk factors drift with the risk-


free short rate — and not with any other market rate involving some kind of risk premium


over the risk-free short rate.


A Simple Example


Consider a simple economy at the dates today and tomorrow with a risky asset, a “stock,”


and a riskless asset, a “bond.” The bond costs 10 USD today and pays off 10 USD


tomorrow (zero interest rates). The stock costs 10 USD today and, with a probability of


60% and 40%, respectively, pays off 20 USD and 0 USD tomorrow. The riskless return of


the bond is 0. The expected return of the stock is , or 20%. This is


the risk premium the stock pays for its riskiness.


Consider now a call option with strike price of 15 USD. What is the fair value of such a


contingent claim that pays 5 USD with 60% probability and 0 USD otherwise? We can


take the expectation, for example, and discount the resulting value back (here with zero


interest rates). This approach yields a value of 0.6 · 5 = 3 USD, since the option pays 5


USD in the case where the stock price moves up to 20 USD and 0 USD otherwise.


However, there is another approach that has been successfully applied to option pricing


problems like this: replication of the option’s payoff through a portfolio of traded


securities. It is easily verified that buying 0.25 of the stock perfectly replicates the option’s


payoff (in the 60% case we then have 0.25 · 20 = 5 USD). A quarter of the stock only


costs 2.5 USD and not 3 USD. Taking expectations under the real-world probability


measure overvalues the option.


Why is this case? The real-world measure implies a risk premium of 20% for the stock


since the risk involved in the stock (gaining 100% or losing 100%) is “real” in the sense


that it cannot be diversified or hedged away. On the other hand, there is a portfolio


available that replicates the option’s payoff without any risk. This also implies that


someone writing (selling) such an option can completely hedge away any risk.


[ 62 ]

Such a


perfectly hedged portfolio of an option and a hedge position must yield the riskless rate in


order to avoid arbitrage opportunities (i.e., the opportunity to make some money out of no


money with a positive probability).


Can we save the approach of taking expectations to value the call option? Yes, we can. We


“only” have to change the probability in such a way that the risky asset, the stock, drifts


with the riskless short rate of zero. Obviously, a (martingale) measure giving equal mass


of 50% to both scenarios accomplishes this; the calculation is . Now,


taking expectations of the option’s payoff under the new martingale measure yields the


correct (arbitrage-free) fair value: 0.5 · 5 + 0.5 · 0 = 2.5 USD.


The General Results

Free download pdf