Topics in Quantitative Finance

Taylor’s formula and Taylor series

Gonzalo Fuentes, CQF

June 2025

1 Taylor’s formula for univariate functions

Consider f (x) a continuous function and we can approximate this function around a point x0 on the real
axis, by the following polynomial of order n, called the Taylor Polynomial of order n:

df 1 dn f n
Pn = f (x0 ) + (x0 ) (x − x0 ) + · · · + (x0 ) (x − x0 ) (1)
dx n! dxn

equation (1) can be written more compactly as:

n k
X (x − x0 )
Pn (x) = f k (x0 ) (2)
k!
k=0

To decide the order of the approximation, Pn (x), the following issues need to be investigated:

• Convergence: when does Pn (x) converge to f (x) as n → ∞.

• Order of approximation: how well does the Taylor polynomial, Pn (x) approximate to f (x)

Theorem 1

Let f (x) be a continuous function which is n + 1 times differentiable, and the derivative of order n + 1
exists and is continuous. Let f (x) − Pn (x) be the approximation error of f (x) by the n-th order Taylor
polynomial, Pn (x). Furthermore, exists a point α0 between x0 and x such that f (x) − Pn (x) can be
expressed as:
n+1
(x − x0 )
f (x) − Pn (x) = f (n+1) (α0 ) (3)
(n + 1)!

Equation (3) can be expressed as a defined integral in the interval [x0 , x], this is:

x
(x − τ ) (n+1)
Z
f (x) − Pn (x) = f (τ ) dτ (4)
x0 n!

1
Theorem 2

Let f (x) be a function which is n + 1 time differentiable, and assume that the n + 1 derivative is a
continuous functions and exists. Then, the approximation of f (x) by the n-th order Taylor polynomial
is of order n + 1 and is expressed as:

1 ′′ 2 1 n

n+1

f (x) = f (x0 ) + f ′ (x0 ) (x − x0 ) + f (x0 ) (x − x0 ) + · · · + f n (x0 ) (x − x0 ) + O (x − x0 ) (5)
2! n!

as x → x0 .

2 Taylor’s formula for multivariate functions

Scalar Functions

Let f : Rn → R be a function of n variables denoted by x = (x1 , · · · , xn ), and let x̂ = (x̂1 , · · · , x̂n ) ∈ Rn .

The linear Taylor expansion of the function f (x) around the point x̂,

n
X ∂f
f (x) ≈ f (x̂) + (xj − x̂j ) (x̂) (6)
j=1
∂xj

is a second order approximation, in the sense that:

n
X ∂f
(x̂) + O ||x − x̂||2


f (x) = f (a) + (xj − x̂j ) (7)
j=1
∂xj

as x → x̂, if all the partial derivatives of second order of f (x) are continuous. Furthermore,


O ||x − x̂||2 can be expressed as:

n

X
O ||x − x̂||2 = |xj − x̂2j |


(8)
j=1

The quadratic Taylor expansion of the function f (x) around the point x̂ is:

n
X ∂f X (xi − x̂i ) (xj − x̂j ) ∂ 2 f
f (x) ≈= f (x̂) + (xj − x̂j ) (x̂) + (x̂) (9)
j=1
∂xj 2 ∂xi ∂xj
i≤i,j≤n

in a third order approximation, in the sense that:

n n
X ∂f X (xi − x̂i ) (xj − x̂j ) ∂ 2 f X
O |xj − x̂j |3 (10)


f (x) ≈= f (x̂) + (xj − x̂j ) (x̂) + (x̂) +
j=1
∂xj 2 ∂xi ∂xj j=1
i≤i,j≤n

as x → x̂, if all the third order partial derivatives are continuous. The linear and quadratic Taylor
expansion, equation (6) and equation (10) can be written using matrix notation as follows:

2
 The gradient vector, Df (x) of the function f (x) is a row vector of size n.
 
∂f ∂f ∂f
Df (x) = (x) ··· (11)
∂x1 ∂x2 ∂xn

Then the Hessian matrix, D2 f (x), is define as a n × n matrix:

′
Df (x) Df (x) = D2 f (x) (12)

where the leading diagonal contains the second order derivatives of f (x) with respect to xj , thus:

∂2f ∂2f ∂2f ∂2f

 
2 ∂x2 ∂x1 ∂x3 ∂x1 ··· ∂xn ∂x1
 ∂x2 1 
 ∂ f ∂2f ∂2f ∂2f 
 ∂x1 ∂x2 ∂x22 ∂x3 ∂x2 ··· ∂xn ∂x2 
D2 f (x) = 

.. .. .. .. ..

 (13)
 . . . . . 
 
∂2f ∂2f ∂2f ∂2f
∂x1 ∂xn ∂x2 ∂xn ∂x3 ∂xn ··· ∂x2n

and let:  
x1 − x̂1
 
 x2 − x̂2 
 
 
x − x̂ =  x3 − x̂3  (14)
 
..
 
 

 . 

xn − x̂n

The linear Taylor expansion of f (x) around the point x̂ can be written in terms of the gradient Df (x):

f (x) ≈ f (x̂) + Df (x) (x − x̂) (15)

f (x) ≈ f (x̂) + Df (x) (x − x̂) + O ||x − x̂||2 , as x → x̂

(16)

Moreover, the quadratic Taylor expansion of f (x) around the point x̂ can be written in terms of
Df (x) and D2 f (x):

1 ′
f (x) ≈ f (x̂) + Df (x̂) (x − x̂) + (x − x̂) D2 f (x) (x − x̂) (17)
2

n
1 ′
X
(x − x̂) D2 f (x) (x − x̂) + O ||xj − x̂j ||3 , as x → x̂


f (x) ≈ f (x̂) + Df (x̂) (x − x̂) + (18)
2 j=1

′ ′
From equation (18) note that (x − x̂) = (x1 − x̂1 , · · · , xn − x̂n ) is a row vector, the transpose of the
column vector from equation (14).

3
3 Taylor’s formula for functions of two variables

Scalar Functions

Let f : R2 → R be a function of two variables. The linear Taylor expansion of f (x, y) around the point
(x̂, ŷ) ∈ R2 .
∂f ∂f
f (x, y) ≈ f (x̂, ŷ) + (x − x̂) (x̂, ŷ) + (y − ŷ) (x̂, ŷ) (19)
∂x ∂y

is a second order approximation, in other words:

∂f ∂f
(x̂, ŷ) + O |x − x̂|2 + O |y − ŷ|2



f (x, y) ≈ f (x̂, ŷ) + (x − x̂) (x̂, ŷ) + (y − ŷ) (20)
∂x ∂y

as (x, y) → (x̂, ŷ), is the second order partial derivatives and cross derivatives are continuous. Then
the quadratic Taylor expansion of f (x, y) around the point (x̂, ŷ) ∈ R2 ,

∂f ∂f 1 ∂2f 1 ∂2f ∂f
f (x, y) ≈ f (x̂, ŷ)+ (x̂, ŷ) ∆x+ (x̂, ŷ) ∆y+ 2
(x̂, ŷ) ∆x2 + (x̂, ŷ) ∆y 2 + (x̂, ŷ) ∆x∆y
∂x ∂y 2! ∂x 2! ∂y 2 ∂x∂y
(21)
Then equation (21) can be expressed in the matrix forms, thus:
 
x − x̂
f (x, y) ≈ f (x̂, ŷ) + Df (x̂, ŷ)   (22)
y − ŷ

 
x − x̂
 + O |x − x̂|2 + O |y − ŷ|2



f (x, y) ≈ f (x̂, ŷ) + Df (x̂, ŷ)  (23)
y − ŷ

as (x, y) → (x̂, ŷ), where Df (x, y) is the gradient vector. The matrix form of the quadratic Taylor
expansion can be expressed as:
   ′  
x − x̂ 1 x − x̂ x − x̂
f (x, y) ≈ f (x̂, ŷ) + Df (x̂, ŷ)  +   D2 f (x̂, ŷ)   (24)
y − ŷ 2 y − ŷ y − ŷ

from equation (24), D2 f (x̂, ŷ) is the Hessian matrix.

4 Financial Applications

Black-Scholes and the Greeks

Let V (St , t) be the value at time t of a European plain vanilla option on an specified underlying asset
with spot price S (t). Recall that V (St , t) is an infinitely many times differentiable function in both
arguments.
To write the quadratic Taylor expansion of V (St , t) around the point (St , t). From equation (21) the

4
Taylor approximation is given by:

∂V ∂V 1 ∂2V 2 1 ∂2V 2 ∂2V

V (St + dSt , t + dt) = V (St , t) + dSt + dt + (dSt ) + (dt) + dSt dt (25)
∂St ∂t 2 ∂St2 2 ∂t2 ∂St ∂t

Then from equation (25), dV (St , t) = V (St + dSt , t + dt) − V (St , t). Furthermore, the evolution of
the price of the underlying asset converge to a lognormal distribution, then the following condition must
be satisfied:
dSt2 ≈ σ 2 St2 dt (26)

Assuming that dt2 = 0 and dSt = µSt dt + σSt dWt , then equation (25) can be expressed as:

∂V ∂V 1 ∂2V 2 2 ∂2V
dV (St , t) = dSt + dt + 2 σ St dt + (µSt dt + σSt dWt ) dt (27)
∂St ∂t 2 ∂St ∂St ∂t

From equation (27), the last term is equal to zero, since dt2 = 0 and dtdWt = 0. Finally the Black-
Scholes equation can be represented as:

∂V ∂V 1 ∂2V
dV (St , t) = dSt + dt + σ 2 St2 2 dt (28)
∂St ∂t 2 ∂St

Also, it is natural to ask how the value of the portfolio changes from t to t + dt. The change in the
portfolio value is due partly to the change in the option value and partly to the change in the specified
underlying asset:
dΠ = dV (St , t) − ∆dSt (29)

∂V ∂V 1 ∂2V
dΠ = dSt + dt + σ 2 St2 2 dt − ∆dSt (30)
∂St ∂t 2 ∂St

The right hand side of equation (30) contains two components: the deterministic are those with the
dt and the random term are those with the dSt . These random terms are the risk in our portfolio, but
¿Is it possible to hedge the risk of the portfolio?. In the theory (and even in the practice), this can be
done by carefully choosing ∆. The random terms are:
 
∂V ∂V
− ∆ dSt → =∆ (31)
∂St ∂St

then the randomness or the risk is reduced to zero. The perfect elimination of the risk, by exploiting
correlation between two instruments is called in the literature as Delta hedging. On the other hand,
after choosing carefully the quantity of ∆, the value of the portfolio changes by the amount:

∂2V
 
∂V 1
dΠ = + σ 2 St2 2 dt (32)
∂t 2 ∂St

This equation suggest that the change in the portfolio value is completely riskless. If we have a
completely risk-free change dΠ in the portfolio value Π then it must be the same as the growth we would

5
get if we put the equivalent amount of cash in a risk-free interest-bearing account:

dΠ = rΠdt (33)

this equation is an example of no-arbitrage principle. Finally, replacing equation (32) in equation
(33), and with the definition that Π = V (St , t) − ∆St , the Black-Scholes equation is given by:

∂2V
 
∂V 1
+ σ 2 St2 2 dt = r (V (St , t) − ∆St ) dt (34)
∂t 2 ∂St

∂2V
   
∂V 1 ∂V
+ σ 2 St2 2 dt = r V (St , t) − St dt (35)
∂t 2 ∂St ∂St

∂V 1 ∂2V ∂V
+ σ 2 St2 2 − rV (St , t) + r St = 0 (36)
∂t 2 ∂St ∂St

Recall that:
∂V ∂2V ∂V
∆= , Γ= and Θ = (37)
∂St ∂St2 ∂t

As a result, the delta (∆) of an option or a portfolio of options is the sensitivity of the option or
portfolio to the underlying. The gamma (Γ), is the second derivative of the position with respect to
the underlying. In other words, gamma is the sensitivity of the delta to the underlying it is a measure
of how much or how often a position must be rehedged in order to maintain a delta-neutral position.
Finally, the theta (Θ) is the rate of change of the option price with time.

References
[1] Stefanica, D. (2011). A primer for the mathematics of financial engineering (2nd ed.), Chapter 2.

[2] Neftci, S. N. (2000). An Introduction to the Mathematics of Financial Derivatives (2nd ed.), Chapter
3.

[3] Wilmott, P. (2007). Paul Wilmott introduces to quantitative finance (2nd ed.), Chapter 6 and 8.