CHAPTER 1.

MATHEMATICAL AND FINANCIAL FRAMEWORK

Example. As an illustrative example, consider an asset price process (S(t))tœ[0,T ] satisfying

dS(t) = µS(t) dt + ‡S(t) dW (t), S(0) > 0,
and consider the function f (x) = log x. Then
1 1
f Õ (x) = , f ÕÕ (x) = ≠ .
x x2
In Itô’s formula, the drift and diffusion coefficients are
a(t) = µS(t), b(t) = ‡S(t),
so that b2 (t) = ‡ 2 S 2 (t). Applying Itô’s formula gives
d log S(t) = f Õ (S(t)) dS(t) + 12 f ÕÕ (S(t)) b2 (t) dt
3 4
1 1 2 1
= µS(t) dt + ‡S(t) dW (t) + 12 ≠ 2 ‡ 2 S 2 (t) dt
S(t) S (t)
= µ dt + ‡ dW (t) ≠ 12 ‡ 2 dt
! "
= µ ≠ 12 ‡ 2 dt + ‡ dW (t).
Integrating over [0, t] yields
! "
log S(t) ≠ log S(0) = µ ≠ 12 ‡ 2 t + ‡W (t),
and therefore !! " "
S(t) = S(0) exp µ ≠ 12 ‡ 2 t + ‡W (t) , t œ [0, T ].
This recovers the explicit representation of the geometric Brownian motion introduced earlier.

The notions introduced in this section provide the minimal stochastic calculus framework required
for the analysis of the Black-Scholes model.

1.3 The Black–Scholes Model

The Black–Scholes model was developed in the early 1970s by Fischer Black and Myron Scholes
and subsequently formalized by Robert C. Merton. It introduced a continuous-time framework for
option pricing based on no-arbitrage arguments and stochastic calculus and remains a foundational
model in mathematical finance.

The model is defined on a filtered probability space ( , F, (Ft )tœ[0,T ] , P) supporting a Brownian
motion (W (t))tœ[0,T ] and is based on the following assumptions:
• There are two traded assets:
– a risk-free asset B(t) satisfying
dB(t) = rB(t) dt, B(0) = 1,
where r is a constant interest rate, so B(t) = ert B(0).
– a risky asset S(t) whose price satisfies
dS(t) = µS(t) dt + ‡S(t) dW (t),
where µ œ R and ‡ > 0 are constants.
• Trading takes place continuously in time over the interval [0, T ].
• The market is frictionless: there are no transaction costs or taxes.
• Assets are perfectly divisible, and short selling is allowed.
• There are no arbitrage opportunities.

8
 1.3. THE BLACK–SCHOLES MODEL

No-arbitrage principle
The no-arbitrage principle rules out trading strategies that generate riskless profits with zero initial
investment. In the Black–Scholes model, this assumption ensures internal price consistency and
underlies the pricing of derivatives.
Informally, a trading strategy is said to be self-financing if changes in its value are due solely to gains
and losses from the assets held, with no additional infusion or withdrawal of capital. An arbitrage
opportunity is a self-financing strategy with zero initial value that produces a non-negative payoff
at maturity, with strictly positive probability of a positive payoff.
A simple example helps clarify this. Assume there are two traded assets A and B with the same
payoff at maturity T , i.e.
A(T ) = B(T ) a.s.
Suppose, however, that their prices at time 0 are different and, without loss of generality,

A(0) < B(0).

Consider the following strategy:

• buy one unit of A at time 0;
• sell one unit of B short at time 0.
It is possible to sell one unit of B because short selling is allowed in Black–Scholes.
Short selling consists in selling an asset that is not currently owned, with the obligation to
repurchase it at a later time. In practice, this is interpreted as borrowing the asset, selling it
on the market, and later buying it back to return it to the lender.
The initial value of the strategy is

V (0) = A(0) ≠ B(0) < 0,

so it produces a positive cash inflow of B(0) ≠ A(0) > 0 at time 0. Invest this amount in the
risk-free asset at rate r, so the time-T value of this cash inflow is
! "
B(0) ≠ A(0) erT .

At maturity, the long and short positions cancel because A(T ) = B(T ), hence the portfolio payoff
at time T is
! " ! "
V (T ) = A(T ) ≠ B(T ) + B(0) ≠ A(0) erT = B(0) ≠ A(0) erT > 0.

Therefore, the strategy requires no net investment and yields a strictly positive payoff with
certainty, which is an arbitrage opportunity. The no-arbitrage condition asserts that no such
strategy exists by requiring that assets with the same future payoffs must have the same current
price.
This example motivates the following formal definition of arbitrage in continuous-time models.

Formal no-arbitrage condition

A trading strategy is a pair of predictable processes

(„(t), Â(t))tœ[0,T ] ,

where „(t) and Â(t) denote the number of units held in the risky asset S(t) and the risk-free asset
B(t), respectively. The associated wealth process is

V (t) = „(t)S(t) + Â(t)B(t).

The strategy is said to be self-financing if, in discounted terms,

dVÂ (t) = „(t) dS(t),

Â

9
CHAPTER 1. MATHEMATICAL AND FINANCIAL FRAMEWORK

where
VÂ (t) := e≠rt V (t), Â := e≠rt S(t).
S(t)

A self-financing strategy is called admissible if its discounted wealth process is almost surely
bounded from below, i.e. there exists a constant C œ R such that

V (t) Ø C for all t œ [0, T ] almost surely.

The role of this admissibility condition is clarified below. An arbitrage opportunity is an admissible
self-financing strategy such that
! " ! "
V (0) = 0, P V (T ) Ø 0 = 1, P V (T ) > 0 > 0.

The market is said to satisfy the no-arbitrage condition if no such strategy exists.

The admissibility condition excludes trading strategies that require arbitrarily large losses before
maturity. In continuous time, one can construct self-financing strategies that drive the wealth
process to ≠Œ at intermediate times while still ending with a non-negative payoff. Such strategies
would require unlimited borrowing and are not economically realistic.

The no-arbitrage condition has a fundamental probabilistic implication. In the Black–Scholes

model, absence of arbitrage is equivalent to the existence of an equivalent probability measure
under which discounted asset prices are martingales. This equivalence is motivated by the following
two theorems. While the first theorem links no-arbitrage to the existence of a risk-neutral measure,
the second addresses when this measure is unique, which is crucial for pricing. This measure is
called an equivalent martingale measure (EM M ) or risk-neutral measure.

Definition 1.9 (Risk-neutral probability measure).

A probability measure Q on ( , F) is called risk-neutral if
1. Q is equivalent to P, that is, for every A œ F,

P(A) = 0 if and only if Q(A) = 0;

2. under Q, the discounted stock price process

 := e≠rt S(t)
S(t)

is a martingale with respect to the filtration (Ft )tœ[0,T ] .

Theorem 1.2 (First Fundamental Theorem of Asset Pricing (FFTAP) [ShreveII]).

If a market model admits a risk-neutral probability measure, then it does not admit arbitrage.

Definition 1.10 (Market completeness).

A market model is said to be complete if for every derivative security with maturity T and payoff X,
there exists an admissible self-financing trading strategy whose terminal wealth equals the payoff,
i.e.
V (T ) = X almost surely.
In this case, the derivative security is said to be hedged or replicated by the trading strategy.

Theorem 1.3 (Second Fundamental Theorem of Asset Pricing (SFTAP) [ShreveII]).

Assume that the market model admits a risk-neutral probability measure. Then the market is
complete if and only if the risk-neutral probability measure is unique.

10
 1.3. THE BLACK–SCHOLES MODEL

Risk-neutral measure in the Black–Scholes model

We now explicitly construct the equivalent martingale measure in the Black–Scholes model. This
will allow us to verify no-arbitrage, prove completeness, and deduce uniqueness.

Construction
Theorem 1.4 (Girsanov’s Theorem).
Let (W (t))tœ[0,T ] be a Brownian motion under P, and let ◊ be an adapted process satisfying
C A ⁄ BD
T
1 2
EP exp 2 ◊(t) dt < Œ.
0

Define the process 3 ⁄ t ⁄ 4

t
1 2
Z(t) := exp ≠ ◊(u) dW (u) ≠ 2 ◊(u) du ,
0 0

and a probability measure Q on ( , FT ) by

dQ --
- = Z(T ).
dP FT
Then Q is equivalent to P, and the process
⁄ t
W (t) := W (t) +
Q
◊(u) du
0

is a Brownian motion under Q with respect to (Ft ).

Remark 1.1. The above integrability condition (often referred to as Novikov’s condition) guarantees
that the exponential process Z is a true martingale with unit expectation.

Girsanov’s theorem characterizes all probability measures Q equivalent to P under which the driving
Brownian motion acquires a drift. Imposing the risk-neutral condition that the discounted stock
price S(t)
 be a martingale under Q uniquely determines this drift. Since the Black–Scholes model
is driven by a single Brownian motion, this condition selects a unique risk-neutral probability
measure as follows
Define
µ≠r dQ -- ! "
◊ := , - = exp ≠◊W (T ) ≠ 12 ◊2 T .
‡ dP FT
By Girsanov’s theorem, the process

W Q (t) := W (t) + ◊t

is a Brownian motion under Q, so that

dW (t) = dW Q (t) ≠ ◊ dt.

Starting from the P-dynamics of the stock price,

dS(t) = µS(t) dt + ‡S(t) dW (t),

we substitute for dW (t) to obtain

! "
dS(t) = µ ≠ ‡◊ S(t) dt + ‡S(t) dW Q (t).

Since ◊ = (µ ≠ r)/‡, this simplifies to

dS(t) = rS(t) dt + ‡S(t) dW Q (t).

11
CHAPTER 1. MATHEMATICAL AND FINANCIAL FRAMEWORK

Define the discounted stock price S(t)

 := e≠rt S(t). Applying the product rule yields

dS(t)
 = e≠rt dS(t) ≠ re≠rt S(t) dt = ‡ S(t)
 dW Q (t).

Thus SÂ has zero drift under Q and is therefore a Q-local martingale. Since
! "
 = S(0) exp ≠ 1 ‡ 2 t + ‡W Q (t) ,
S(t) 2

it is integrable and hence a true Q-martingale.

Therefore, Q is a martingale (risk–neutral) probability measure.
Remark 1.2. By the first fundamental theorem of asset pricing the existence of an equivalent
martingale measure ensures that the market is no-arbitrage.

Completeness
Theorem 1.5 (Martingale Representation Theorem (MRP)).
Let ( , F, (Ft )tœ[0,T ] , Q) be a filtered probability space satisfying the usual conditions, where (Ft ) is
generated by a one-dimensional Brownian motion W Q . If M (t) is a square-integrable Q-martingale,
then there exists a predictable process h(t) with
C⁄ D
T
EQ h(t)2 dt < Œ
0

such that ⁄ t
M (t) = M (0) + h(u) dW Q (u), t œ [0, T ].
0

The process h is unique up to indistinguishability.

Having fixed the risk-neutral measure via Girsanov’s theorem, the martingale representation
theorem ensures that every payoff can be replicated. The next proposition is fundamental is
also used to prove completeness.

Proposition 1.2.
Let X œ L2 (Q) be FT -measurable and define, for t œ [0, T ],
# $
M (t) := EQ e≠rT X | Ft .

Then (M (t))tœ[0,T ] is a Q-martingale with respect to (Ft ).

Proof. First, M (t) is Ft -measurable by definition of conditional expectation. Moreover, since

e≠rT X œ L2 (Q), the conditional expectation is integrable and EQ [|M (t)|] < Œ for every t œ [0, T ].
Let 0 Æ s Æ t Æ T . Using the tower property (iterated conditioning) of conditional expectation,
Ë # $ -- È # $
EQ [M (t) | Fs ] = EQ EQ e≠rT X | Ft - Fs = EQ e≠rT X | Fs = M (s).

Therefore (M (t)) satisfies the martingale property, hence it is a Q-martingale.

To prove completeness, one has to show that for every derivative security with maturity T and
payoff X, there exists an admissible self-financing strategy such that V (T ) = X almost surely.

Let X be an FT -measurable random variable representing the payoff at time T . By the above
proposition, the discounted value process
# $
M (t) = EQ e≠rT X | Ft

12
 1.3. THE BLACK–SCHOLES MODEL

is a Q-martingale. From the martingale representation theorem there exists a predictable process
h(t) such that
⁄ t
M (t) = M (0) + h(u) dW Q (u), t œ [0, T ].
0
Under the risk-neutral measure Q, the discounted stock price satisfies

dS(t)
 = ‡ S(t)
 dW Q (t).

Consequently, stochastic integrals with respect to W Q can be expressed as stochastic integrals with
respect to S.
 Indeed, defining
h(t)
„(t) := ,
Â
‡ S(t)
we obtain ⁄ ⁄
t t
h(u) dW Q (u) = „(u) dS(u).
Â
0 0

Therefore, ⁄ t
M (t) = M (0) + „(u) dS(u),
Â
0
which shows that the discounted value process evolves solely through gains and losses from holding
„(t) units of the stock. In particular, this corresponds to a self-financing trading strategy whose
discounted wealth equals M (t).
Since M (T ) = e≠rT X, the corresponding undiscounted wealth satisfies V (T ) = X almost surely.
Thus the payoff X is exactly replicated, and every derivative security can be hedged. Hence the
Black–Scholes market is complete.

Uniqueness
Uniqueness follows from the second fundamental theorem of asset pricing (SFTAP). Assuming an
EMM exists, completeness implies that the risk-neutral measure is unique.
The next diagram summarizes existence and uniqueness of the risk-neutral measure.

MRP SFTAP
Existence of an EMM (Girsanov) ===∆ Completeness =====∆ Uniqueness of the EMM.

Risk-neutral valuation
Let X be the payoff of a derivative security at maturity T , with X œ L1 (Q). Since the market is
complete, there exists a unique arbitrage-free price process V (t), given by

V (t) = e≠r(T ≠t) EQ [X | Ft ] , t œ [0, T ].

13