risk neutral valuation - christian frieschristian-fries.de/finmath/lecture12/fries -...
TRANSCRIPT
![Page 1: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/1.jpg)
Risk Neutral Valuation
Christian Fries
Version 2.2
http://www.christian-fries.de/finmath
April 19-20, 2012
© copyright 2012 Christian Fries 1 / 51
![Page 2: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/2.jpg)
Outline
NotationDifferential EquationsMonte-Carlo Simulation
Modeling
Risk Neutral ValuationChange of Measure / DriftExample: Black Scholes ModelExample: Black-Scholes Model Monte-Carlo Option Pricer
© copyright 2012 Christian Fries 2 / 51
![Page 3: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/3.jpg)
NOTATIONDIFFERENTIAL EQUATIONS
![Page 4: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/4.jpg)
Notation I
Integral: ∫ tn
0f (t) dt ≈
n−1
∑i=0
f (ti) ·∆ti
Discrete Interpretation:I Integral is approximately the sum of f (ti) ·∆ti .
I For piecewise constant function f , constant on [ti , ti+1)we have "‘="’ above.
© copyright 2012 Christian Fries 4 / 51
![Page 5: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/5.jpg)
Notation II
Differential Equation: Define the function t 7→ X (t) by specifying itschange:
dX = f (t) dt :⇔ X (T ) = X (0) +∫ T
0f (t) dt
Discrete Interpretation:I f is the change of X per unit time (rate of change):
∆X (ti) = f (ti) ·∆ti ⇔ ∆X (ti)/
∆ti = f (ti).
© copyright 2012 Christian Fries 5 / 51
![Page 6: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/6.jpg)
Notation III
Differential Equation - Special Case: Define the function X byspecifying its relative change.
dX = f (t) ·X (t) dt ⇔ X (T ) = X (0) +∫ T
0f (t) ·X (t) dt
Discrete Interpretation:I f is the relative change of X per time (percentage rate of change):
∆X (ti) = f (ti) ·X (ti) ·∆ti ⇔ ∆X(ti )X(ti )
/∆ti = f (ti).
© copyright 2012 Christian Fries 6 / 51
![Page 7: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/7.jpg)
Notation IV
Exercise: Excel Sheet with Discretization of Differential Equation
© copyright 2012 Christian Fries 7 / 51
![Page 8: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/8.jpg)
Notation V
State, Pobability, Pobability Measure:
Ω = ω1,ω2, . . . ,ωn (probability space)P(ωi) (probability that we are in state ωi )P (probability measure)
Probability of an State Configuration (Event):
P(ω1,ω2, . . . ,ωk) = P(ω1) + P(ω2) + . . .+ P(ωk)
Random Variable:
X : Ω→ IR Example: X (ωi) payment that depends on the state ωi .
Expectation:EP(X ) := ∑
ωi∈Ω
X (ωi) ·P(ωi)
© copyright 2012 Christian Fries 8 / 51
![Page 9: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/9.jpg)
Notation VI
Conditional Expectation:
EP(X |F ) :=1
P(F ) ∑ωi∈F
X (ωi) ·P(ωi)
© copyright 2012 Christian Fries 9 / 51
![Page 10: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/10.jpg)
NOTATIONMONTE-CARLO SIMULATION
![Page 11: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/11.jpg)
Notation VIMonte-Carlo Simulation
Expectation:EP(X ) := ∑
ωi∈Ω
X (ωi) ·P(ωi)
Monte-Carlo Simulation: Numerical Approximation of Expectation:Let
Ω := ω1, ω2, . . . , ωm
denote elements from Ω - a drawing from Ω, i.e. a set of samples. Someωi ’s may be the same and we may have more ωi ’s than Ω has elements.Our set of sample paths should have the following property:
A path ω ∈ Ω occures in Ω approximately m ·P(ω) times.
Then we have
EP(X ) := ∑ωi∈Ω
X (ωi) ·P(ωi)≈1m ∑
ωi∈Ω
X (ωi)
© copyright 2012 Christian Fries 11 / 51
![Page 12: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/12.jpg)
Outline
NotationDifferential EquationsMonte-Carlo Simulation
Modeling
Risk Neutral ValuationChange of Measure / DriftExample: Black Scholes ModelExample: Black-Scholes Model Monte-Carlo Option Pricer
© copyright 2012 Christian Fries 12 / 51
![Page 13: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/13.jpg)
ModelingRandom Variable and Stochastic Processes
Financial Product: A stream of payments depending on eventsdescribed in a contract.
Modeling:Value / Payment→ Random Variable
X : Ω→ IR.
I Value of an asset (aka. underlying) at a fixed time t dep. on thestates of the world.
I Payment of a financial product at a fixed time t dep. on state ofunderlying assets.
I Value of the financial product at a fixed time t depending on statesof the assets.
© copyright 2012 Christian Fries 13 / 51
![Page 14: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/14.jpg)
ModelingRandom Variable and Stochastic Processes
Modeling:Value Process / Payoff Stream→ Stochastic Process:= family of random variables over time
S : [0,∞)×Ω→ IR
S(ω) with ωεΩ is a map [0,∞)→ IR: the path of S in state ω. Note: Allrandom variables (at all times t) are given over the same space (Ω,F )
ω ∈ Ω ↔ path / history / chain of events
© copyright 2012 Christian Fries 14 / 51
![Page 15: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/15.jpg)
ModelingRandom Variable and Stochastic Processes
Modeling:Information is modeled through the filtration:
I Family of σ algebras Ft .I Ft ⊂Fs, t < s.I Elements of Ft are the events, which may be known at time t .
Measurable:
X is FT measurable ↔ X is known in time T
Natural Condition (for stochastic process):
S(T ) is FT measurable.
© copyright 2012 Christian Fries 15 / 51
![Page 16: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/16.jpg)
ModelingRandom Variable and Stochastic Processes
Modeling:Information is modeled through the filtration:
I Family of σ algebras Ft .I Ft ⊂Fs, t < s.I Elements of Ft are the events, which may be known at time t .
Measurable:
X is FT measurable ↔ X is known in time T
Natural Condition (for stochastic process):
S(T ) is FT measurable.
© copyright 2012 Christian Fries 15 / 51
![Page 17: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/17.jpg)
ModelingRandom Variable and Stochastic Processes
Modeling:Information is modeled through the filtration:
I Family of σ algebras Ft .I Ft ⊂Fs, t < s.I Elements of Ft are the events, which may be known at time t .
Measurable:
X is FT measurable ↔ X is known in time T
Natural Condition (for stochastic process):
S(T ) is FT measurable.
© copyright 2012 Christian Fries 15 / 51
![Page 18: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/18.jpg)
ModelingRandom Variable and Stochastic Processes
Example: (Time discrete) stochastic process of coin toss at T1, T2, T3.Modeling (e.g. a bet):
Ω = (h,h,h),(h,h, t),(h, t ,h), . . . ,(t , t , t) - Prob’ space (head or tail).
X : Ω 7→ IR - Bet with a single payoff
S : T1,T2,T3×Ω 7→ IR
S(Tk ) paid at time Tk .
-Bet with paymentsor evolution of value(stochastic process)
Natural Condition (*): S(Tk ) may depend on events known on or beforeTk only.
⇔ S(Tk ,(e1, . . . ,en)) = const. ∀ei ∈ h, t mit i > k
© copyright 2012 Christian Fries 16 / 51
![Page 19: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/19.jpg)
ModelingRandom Variable and Stochastic Processes
Example: (Time discrete) stochastic process of coin toss at T1, T2, T3.Modeling (e.g. a bet):
Ω = (h,h,h),(h,h, t),(h, t ,h), . . . ,(t , t , t) - Prob’ space (head or tail).
X : Ω 7→ IR - Bet with a single payoff
S : T1,T2,T3×Ω 7→ IR
S(Tk ) paid at time Tk .
-Bet with paymentsor evolution of value(stochastic process)
Natural Condition (*): S(Tk ) may depend on events known on or beforeTk only.
⇔ S(Tk ,(e1, . . . ,en)) = const. ∀ei ∈ h, t mit i > k
© copyright 2012 Christian Fries 16 / 51
![Page 20: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/20.jpg)
ModelingRandom Variable and Stochastic Processes
Example: (Time discrete) stochastic process of coin toss at T1, T2, T3.Define family of σ algebras (filtration):
F0 = 0,ΩF1 = σ((h,∗,∗),(t ,∗,∗))F2 = σ((h,h,∗),(h, t ,∗),(t ,h,∗),(t , t ,∗))F3 = σ((h,h,h),(h,h, t),(h, t ,h), . . . ,(t , t , t))
ω1
ω8
T1 T2 T3T0 T4
ω2h
t
ω4ω5ω6ω7
ω3
Condition (*) ⇔ S(Tk ) is Fk -measurable ⇔: S is adapted to Fk.© copyright 2012 Christian Fries 17 / 51
![Page 21: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/21.jpg)
ModelingRandom Variable and Stochastic Processes
Example: (Time discrete) stochastic process of coin toss at T1, T2, T3.Define family of σ algebras (filtration):
F0 = 0,ΩF1 = σ((h,∗,∗),(t ,∗,∗))F2 = σ((h,h,∗),(h, t ,∗),(t ,h,∗),(t , t ,∗))F3 = σ((h,h,h),(h,h, t),(h, t ,h), . . . ,(t , t , t))
ω1
ω8
T1 T2 T3T0 T4
ω2h
t
ω4ω5ω6ω7
ω3
Condition (*) ⇔ S(Tk ) is Fk -measurable ⇔: S is adapted to Fk.© copyright 2012 Christian Fries 17 / 51
![Page 22: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/22.jpg)
ModelingRandom Variable and Stochastic Processes
Example: (Time discrete) stochastic process of coin toss at T1, T2, T3.Define family of σ algebras (filtration):
F0 = 0,ΩF1 = σ((h,∗,∗),(t ,∗,∗))F2 = σ((h,h,∗),(h, t ,∗),(t ,h,∗),(t , t ,∗))F3 = σ((h,h,h),(h,h, t),(h, t ,h), . . . ,(t , t , t))
ω1
ω8
T1 T2 T3T0 T4
ω2h
t
ω4ω5ω6ω7
ω3
Condition (*) ⇔ S(Tk ) is Fk -measurable ⇔: S is adapted to Fk.© copyright 2012 Christian Fries 17 / 51
![Page 23: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/23.jpg)
ModelingRandom Variable and Stochastic Processes
Example: (Time discrete) stochastic process of coin toss at T1, T2, T3.Define family of σ algebras (filtration):
F0 = 0,ΩF1 = σ((h,∗,∗),(t ,∗,∗))F2 = σ((h,h,∗),(h, t ,∗),(t ,h,∗),(t , t ,∗))F3 = σ((h,h,h),(h,h, t),(h, t ,h), . . . ,(t , t , t))
ω1
ω8
T1 T2 T3T0 T4
ω2h
t
ω4ω5ω6ω7
ω3
Condition (*) ⇔ S(Tk ) is Fk -measurable ⇔: S is adapted to Fk.© copyright 2012 Christian Fries 17 / 51
![Page 24: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/24.jpg)
ModelingRandom Variable and Stochastic Processes
Example: (Time discrete) stochastic process of coin toss at T1, T2, T3.Define family of σ algebras (filtration):
F0 = 0,ΩF1 = σ((h,∗,∗),(t ,∗,∗))F2 = σ((h,h,∗),(h, t ,∗),(t ,h,∗),(t , t ,∗))F3 = σ((h,h,h),(h,h, t),(h, t ,h), . . . ,(t , t , t))
ω1
ω8
T1 T2 T3T0 T4
ω2
ω4ω5ω6ω7
ω3
h
t
Condition (*) ⇔ S(Tk ) is Fk -measurable ⇔: S is adapted to Fk.© copyright 2012 Christian Fries 17 / 51
![Page 25: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/25.jpg)
ModelingRandom Variable and Stochastic Processes
Example: (Time discrete) stochastic process of coin toss at T1, T2, T3.Define family of σ algebras (filtration):
F0 = 0,ΩF1 = σ((h,∗,∗),(t ,∗,∗))F2 = σ((h,h,∗),(h, t ,∗),(t ,h,∗),(t , t ,∗))F3 = σ((h,h,h),(h,h, t),(h, t ,h), . . . ,(t , t , t))
ω1
ω8
T1 T2 T3T0 T4
ω2
ω4ω5ω6ω7
ω3h
t
Condition (*) ⇔ S(Tk ) is Fk -measurable ⇔: S is adapted to Fk.© copyright 2012 Christian Fries 17 / 51
![Page 26: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/26.jpg)
ModelingRandom Variable and Stochastic Processes
Modeling:Prototype of a stochastic process - building block of Itô processes:
Brownian Motion: WI W (t) defined over (Ω,F ,P).I W (0) = 0.I W (t) normal distribution with mean 0 and standard deviation
√t .
I W (t2)−W (t1) normal distribution with mean 0 andstandard deviation
√t2− t1 (i.i.d).
I W (·,ω) continuous (but nowhere differentiable) function [0,∞)→ IR(P-a.s.).
© copyright 2012 Christian Fries 18 / 51
![Page 27: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/27.jpg)
ModelingRandom Variable and Stochastic Processes
Modeling:Prototype of a stochastic process - building block of Itô processes:
Brownian Motion: WI W (t) defined over (Ω,F ,P).I W (0) = 0.I W (t) normal distribution with mean 0 and standard deviation
√t .
I W (t2)−W (t1) normal distribution with mean 0 andstandard deviation
√t2− t1 (i.i.d).
I W (·,ω) continuous (but nowhere differentiable) function [0,∞)→ IR(P-a.s.).
© copyright 2012 Christian Fries 18 / 51
![Page 28: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/28.jpg)
ModelingRandom Variable and Stochastic Processes
Construction of realizations at discrete times:
W (tk ) :=k−1
∑i=0
∆W (ti) (0 = t0 < t1 < .. .),
where
W (t0) := 0 , ∆W (ti) = (W (ti+1)−W (ti))∼N (0,√
ti+1− ti) (i.i.d).
ω - (path)
T1 T2 T3T0 T4
W(t) ΔW(T1 ) = W
(T2 )-W(T1 )
ΔW(T2 ) = W
(T3 )-W(T2 )
ΔW(T3 ) = W
(T4 )-W(T3 )
© copyright 2012 Christian Fries 19 / 51
![Page 29: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/29.jpg)
ModelingRandom Variable and Stochastic Processes
Construction of realizations at discrete times:
W (tk ) :=k−1
∑i=0
∆W (ti) (0 = t0 < t1 < .. .),
where
W (t0) := 0 , ∆W (ti) = (W (ti+1)−W (ti))∼N (0,√
ti+1− ti) (i.i.d).
Infinitesimal Notation:
W (t) =:∫ t
0dW (τ).
© copyright 2012 Christian Fries 19 / 51
![Page 30: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/30.jpg)
ModelingRandom Variable and Stochastic Processes
Stochastic Differential Equation:
dS = µ(t ,S(t)) dt + σ(t ,S(t))dW (t)
Euler Discretization (e.g.):
∆S(ti)︸ ︷︷ ︸S(ti+1)− S(ti)
= µ(ti , S(ti)) · ∆ti︸︷︷︸ti+1− ti
+ σ(ti , S(ti)) · ∆Wi︸︷︷︸∼N (0,
√∆ti)
,
i.e.
S(ti+1) = S(ti) + µ(ti , S(ti)) · (ti+1− ti) + σ(ti , S(ti)) ·∆Wi ,
with S(0) = S(0).
S(ti) | 0 = t0 < t1 < .. . is the Euler discretization of S(t) | t ≥ 0.© copyright 2012 Christian Fries 20 / 51
![Page 31: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/31.jpg)
ModelingRandom Variable and Stochastic Processes
Example: log normal process for stock value S:
dS(t) = µ(t)S(t)dt +σ(t)S(t)dW (t) ⇔ dS(t)S(t)
= µ(t)dt +σ(t)dW (t)
Euler Scheme:
S(tj+1) = S(tj) + µ(tj) · S(tj) ·∆tj + σ(tj) · S(tj) · ∆Wj︸︷︷︸∼N (0,
√∆tj)
,
Log Euler Scheme:
S(tj+1) = S(tj) ·exp(
(µ(tj)−12
σ(tj)2) ·∆tj + σ(tj) · ∆Wj︸︷︷︸∼N (0,
√∆tj)
),
© copyright 2012 Christian Fries 21 / 51
![Page 32: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/32.jpg)
ModelingRandom Variable and Stochastic Processes
Example: log normal process for stock value S:
dS(t) = µ(t)S(t)dt +σ(t)S(t)dW (t) ⇔ dS(t)S(t)
= µ(t)dt +σ(t)dW (t)
Euler Scheme:
S(tj+1) = S(tj) + µ(tj) · S(tj) ·∆tj + σ(tj) · S(tj) · ∆Wj︸︷︷︸∼N (0,
√∆tj)
,
Log Euler Scheme:
S(tj+1) = S(tj) ·exp(
(µ(tj)−12
σ(tj)2) ·∆tj + σ(tj) · ∆Wj︸︷︷︸∼N (0,
√∆tj)
),
© copyright 2012 Christian Fries 21 / 51
![Page 33: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/33.jpg)
ModelingRandom Variable and Stochastic Processes
Example: log normal process for stock value S:
dS(t) = µ(t)S(t)dt +σ(t)S(t)dW (t) ⇔ dS(t)S(t)
= µ(t)dt +σ(t)dW (t)
Euler Scheme:
S(tj+1) = S(tj) + µ(tj) · S(tj) ·∆tj + σ(tj) · S(tj) · ∆Wj︸︷︷︸∼N (0,
√∆tj)
,
Log Euler Scheme:
S(tj+1) = S(tj) ·exp(
(µ(tj)−12
σ(tj)2) ·∆tj + σ(tj) · ∆Wj︸︷︷︸∼N (0,
√∆tj)
),
© copyright 2012 Christian Fries 21 / 51
![Page 34: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/34.jpg)
ModelingRandom Variable and Stochastic Processes
Exercise: Excel Sheet with Monte-Carlo Simulation
© copyright 2012 Christian Fries 22 / 51
![Page 35: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/35.jpg)
Outline
NotationDifferential EquationsMonte-Carlo Simulation
Modeling
Risk Neutral ValuationChange of Measure / DriftExample: Black Scholes ModelExample: Black-Scholes Model Monte-Carlo Option Pricer
© copyright 2012 Christian Fries 23 / 51
![Page 36: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/36.jpg)
Risk Neutral Valuation
Product A: Weather DerivativeAt time T2 > 0 measure amount of rain R(T2) (in mm) fallen at apredetermined place for a predetermined period. Pay the Euro amount
A(T2) := (R(T2)−X ) · Cmm
.
What determines the value A(T1) of this contract at time T1 < T2?Note: R(T2) is stochastic⇒ Value depends on assessment of the probability of rain, risk, . . .Product B: Equity DerivativeAt time T2 measure the quoted value of IBM S(T2) (in C). Pay the Euroamount
B(T2) := (S(T2)−X )
What determined the value B(T1) of this contract at time T1 < T2?⇒ Product looks most similar to the first one.
© copyright 2012 Christian Fries 24 / 51
![Page 37: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/37.jpg)
Risk Neutral Valuation
Replication:The payoff of B can be replicated through products traded in T1:Let P(T2;T1) denote the value of a credit which has to be payed back inT2 by an amount of 1 (→ Zero Bond). In T1:
Buy stock S(T1) and X times a credit P(T2;T1)(stock long, bond short).
Then
Value in T2: S(T2)−X = B(T2)
Value in T1: S(T1)−X ·P(T2;T1)!
= B(T1)
⇒ Value of replication portfolio is a “fair value” for the derivative.
© copyright 2012 Christian Fries 25 / 51
![Page 38: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/38.jpg)
Risk Neutral Valuation
Discrete Time (T1,T2), Two Assets (S,B), Two States (ω1,ω2)Given: In T2 we have two disjoint events (states) ω1 and ω2, the
payoff of a derivative product V (T2,ωi) and two traded as-sets S und B.
Wanted: A portfolio αS + βB, which replicates the value V (T2) inT2 for any state.⇒ Its value in T1 determines the (fair) value V (T1) of V→ cost of replication.
Solution:
α ·S(T1) + β ·B(T1)
α ·S(T2;ω1) + β ·B(T2;ω1)!
= V (T2;ω1)
α ·S(T2;ω2) + β ·B(T2;ω2)!
= V (T2;ω2)
© copyright 2012 Christian Fries 26 / 51
![Page 39: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/39.jpg)
Risk Neutral Valuation
Linear equation, two equations, two unknowns.
α ·S(T1) + β ·B(T1)
α ·S(T2;ω1) + β ·B(T2;ω1)!
= V (T2;ω1)
α ·S(T2;ω2) + β ·B(T2;ω2)!
= V (T2;ω2)
Solvable
⇔ S(T2;ω1) ·B(T2;ω2)−S(T2;ω2) ·B(T2;ω1) 6= 0
B 6=0⇔ S(T2;ω1)
B(T2;ω1)6= S(T2;ω2)
B(T2;ω2)
Overhead: For every derivative product V , the replication port-folio α(t), β (t) has to be calculated (though everytime step t).
© copyright 2012 Christian Fries 27 / 51
![Page 40: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/40.jpg)
Risk Neutral Valuation
Same example again - but consider everything divided by B
α · S(T1)B(T1) + β ·1
α · S(T2;ω1)B(T2;ω1) + β ·1 !
= V (T2;ω1)B(T2;ω1)
α · S(T2;ω2)B(T2;ω2) + β ·1 !
= V (T2;ω2)B(T2;ω2)
p
1−p
© copyright 2012 Christian Fries 28 / 51
![Page 41: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/41.jpg)
Risk Neutral Valuation
Same example again - but consider everything divided by B
α · S(T1)B(T1) + β ·1
α · S(T2;ω1)B(T2;ω1) + β ·1 !
= V (T2;ω1)B(T2;ω1)
α · S(T2;ω2)B(T2;ω2) + β ·1 !
= V (T2;ω2)B(T2;ω2)
p
1−p
In general
V (T1)
B(T1)6= EP
(V (T2)
B(T2)
)and thus
S(T1)
B(T1)6= EP
(S(T2)
B(T2)
).
© copyright 2012 Christian Fries 28 / 51
![Page 42: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/42.jpg)
Risk Neutral Valuation
Same example again - but consider everything divided by B
α · S(T1)B(T1) + β ·1
α · S(T2;ω1)B(T2;ω1) + β ·1 !
= V (T2;ω1)B(T2;ω1)
α · S(T2;ω2)B(T2;ω2) + β ·1 !
= V (T2;ω2)B(T2;ω2)
q
1−q
© copyright 2012 Christian Fries 28 / 51
![Page 43: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/43.jpg)
Risk Neutral Valuation
Same example again - but consider everything divided by B
α · S(T1)B(T1) + β ·1
α · S(T2;ω1)B(T2;ω1) + β ·1 !
= V (T2;ω1)B(T2;ω1)
α · S(T2;ω2)B(T2;ω2) + β ·1 !
= V (T2;ω2)B(T2;ω2)
q
1−q
However, if sign(
S(T2;ω2)B(T2;ω2) −
S(T1)B(T1)
)6= sign
(S(T2;ω1)B(T2;ω1) −
S(T1)B(T1)
)there exists a
prob.-measure Q s.th.
S(T1)
B(T1)= EQ
(S(T2)
B(T2)
)and thus
V (T1)
B(T1)= EQ
(V (T2)
B(T2)
).
© copyright 2012 Christian Fries 28 / 51
![Page 44: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/44.jpg)
Risk Neutral Valuation
Equivalent Martingale MeasureThe measure Q is the so called equivalent martingale measure. It doesnot depend on V (!) - this does not hold if one considers
V (T1)!
= E(V (T2)) in place of V (T1)B(T1)
!= E
(V (T2)B(T2)
).
NuméraireThe measure Q depends on the reference quantity (here B), the socalled numéraire.Here:
q · S(T2;ω1)
B(T2;ω1)+ (1−q) · S(T2;ω2)
B(T2;ω2)!
=S(T1)
B(T1)⇒ q =
S(T2;ω1)B(T2;ω1) −
S(T1)B(T1)
S(T2;ω1)B(T2;ω1) −
S(T2;ω2)B(T2;ω2)
© copyright 2012 Christian Fries 29 / 51
![Page 45: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/45.jpg)
Risk Neutral Valuation
Universal Pricing Theorem:
V (0)
N(0)= EQN
(V (Tn)
N(Tn)
∣∣ F0
)Assume V consists of finite number of payments
Xi paid in Ti , ⇒ N-relative payment value: EQN( Xi
N(Ti)
∣∣F0).
Value
⇒ EQN(
VN∣∣ F0
)=
n
∑i=1
EQN(
Xi
N(Ti)
∣∣ F0
)
© copyright 2012 Christian Fries 30 / 51
![Page 46: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/46.jpg)
Risk Neutral Valuation
Monte-Carlo Simulation:Sample space Ω = ω1, . . . ,ωm, e.g. m ≈ 10000
I V (0)N(0) = EQN (V
N
∣∣ F0)
= ∑ni=1 EQN
(Xi
N(Ti )
∣∣ F0
)I EQN
(Xi
N(Ti )
∣∣F0
)≈ 1
m ∑ω∈ΩXi (ω)
N(Ti ;ω)
⇒ V (0) = N(0) ·EQN(
VN∣∣ F0
)≈ 1
m ∑ω∈Ω
n
∑i=1
Xi(ω) · N(0)
N(Ti ;ω)︸ ︷︷ ︸state price deflator / discount factor
© copyright 2012 Christian Fries 31 / 51
![Page 47: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/47.jpg)
Risk Neutral Valuation
Monte-Carlo SimulationAdvantages
I Pricing of path dependent options in a natural way.I High dimensions possible (but slower convergence in high
dimensions).I Straight forward to implement.
ChallengesI Sensitivities (partial derivatives of option price w.r.t. model
parameters) tend to be unstable (harder to get them stable).I Pricing of Bermudan option needs additional effort.I Calibration.
© copyright 2012 Christian Fries 32 / 51
![Page 48: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/48.jpg)
Risk Neutral Valuation
Monte-Carlo SimulationAlternatives
I Analytic pricing formulas - only for simple models and simplepayoffs.
I Numerical integration - only for simple models and European styleoptions.
I Lattice models (PDE, Tree, Markov Functional) - only for lowdimensional models
© copyright 2012 Christian Fries 33 / 51
![Page 49: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/49.jpg)
RISK NEUTRAL VALUATIONCHANGE OF MEASURE / DRIFT
![Page 50: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/50.jpg)
Change of Drift Trick I
Risk Neutral Pricing: Find a measure Q such that for all underlyingassets S we have
EQ(
S(Ti+1)
N(Ti+1)| FTi
)=
S(Ti)
N(Ti),
then for a derivative the cost of replication V satisfies
EQ(
V (Ti+1)
N(Ti+1)| FTi
)=
V (Ti)
N(Ti).
Martingale Property: (in place of SN or V
N we simple write X )
EQ(X (Ti+1) | FTi
)= X (Ti) ⇔ EQ(X (Ti+1)−X (Ti)︸ ︷︷ ︸
=:∆X(Ti )
| FTi
)= 0
How to calculate the measure Q?
© copyright 2012 Christian Fries 35 / 51
![Page 51: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/51.jpg)
Change of Drift Trick II
Never need to (explicitly) calculate the measure Q:Conditional Expectation:
EQ(∆X (Ti) | FTi ) = ∑ω∈Ω
∆X (Ti ,ω) ·Q(ω)
interpretation 1: the measure has changed from P to Q:
= ∑ω∈Ω
∆X (Ti ,ω) · Q(ω)P(ω)
P(ω)︸ ︷︷ ︸measure changed
interpretation 2: the values have changed:
= ∑ω∈Ω
∆X (Ti ,ω)Q(ω)P(ω)︸ ︷︷ ︸
value changed
·P(ω)
© copyright 2012 Christian Fries 36 / 51
![Page 52: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/52.jpg)
Change of Drift Trick III
Question: How does the process look like, if it has to satisfy themartingale property?
Answer: The drift is zero.
dX (t) = µdt + σdW (t)
and
EQ(X (Ti+1)−X (Ti) | FTi ) = 0
⇒ µ = 0
Note: X (Ti+1)−X (Ti) = ∆X =∫ Ti+1
TidX (t).
© copyright 2012 Christian Fries 37 / 51
![Page 53: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/53.jpg)
Change of Drift Trick IV
Conclusion:Instead of calculating the pricing measure (martingale measure), simplywrite down the process and calculate the correct drift.
change of measure ⇔ change of drift
A Monte-Carlo simulation of that processcorresponds to a simulation under the pricing measure.
© copyright 2012 Christian Fries 38 / 51
![Page 54: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/54.jpg)
Change of Drift Trick V
Itô Lemma: Main Tool to Calculate the Drift of ProcessesLet X denote an Itô prozess with
dX (t) = µ dt + σ dW .
Let g(t ,x) denote some function g ∈ C2([0,∞]×R). The we have that
Y (t) := g(t ,X (t))
is an Itô process with
dYt =∂g∂ t
(t ,Xt ) dt +∂g∂x
(t ,Xt ) dXt +12
∂ 2g∂x2 (t ,Xt ) · (dXt )
2,
where (dXt )2 = (dXt ) · (dXt ) is given by formal expansion with
dt ·dt = 0, dt ·dW = 0,dW ·dt = 0, dW ·dW = dt ,
© copyright 2012 Christian Fries 39 / 51
![Page 55: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/55.jpg)
Change of Drift Trick VI
i.e.
(dXt )2 = (dXt ) · (dXt ) = (µ dt + σ dW ) · (µ dt + σ dW )
= µ2 ·dt ·dt + µ ·σ ·dt ·dW + µ ·σ ·dW ·dt + σ
2 ·dW ·dW
= σ2 dt
© copyright 2012 Christian Fries 40 / 51
![Page 56: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/56.jpg)
Change of Drift Trick VII
Itô Lemma: Example:
dX (t) = µX (t) dt + σX (t) dW .
Consider logarithm of X
Y (t) := log(X (t))
thendY (t) = (µ− 1
2σ
2) dt + σ dW .
© copyright 2012 Christian Fries 41 / 51
![Page 57: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/57.jpg)
RISK NEUTRAL VALUATIONEXAMPLE: BLACK SCHOLES MODEL
![Page 58: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/58.jpg)
Example: Black-Scholes Model
Black-Scholes Model:dS = µSdt + σSdW ⇔ dS
S= µdt + σdW
Interpretation: Consider discrete times t0, t1, t2, t3, . . .. Then:∆S(ti)
S= µ ·∆ti + σ ·∆W (ti)
where∆ti = ti+1− ti A time period.
∆S(ti)S(ti)
=S(ti+1)−S(ti)
S(ti)Relative change of the stock over theperiod ∆ti = ti+1− ti .
µ ·∆ti = µ · (ti+1− ti)Risk less part: A deterministic drift.µ = annualized risk less rate of return.
σ ·∆W (ti) = σ · (W (ti+1)−W (ti))︸ ︷︷ ︸Normal distributedwith variance ∆ti
Risky part: A normal distributed randomvariable. σ = annualized standard devi-ation of the return.
© copyright 2012 Christian Fries 43 / 51
![Page 59: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/59.jpg)
Example: Black-Scholes Model
Black-Scholes Model:dS = µSdt + σSdW ⇔ dS
S= µdt + σdW
Interpretation: Consider discrete times t0, t1, t2, t3, . . .. Then:∆S(ti)
S= µ ·∆ti + σ ·∆W (ti)
where∆ti = ti+1− ti A time period.
∆S(ti)S(ti)
=S(ti+1)−S(ti)
S(ti)Relative change of the stock over theperiod ∆ti = ti+1− ti .
µ ·∆ti = µ · (ti+1− ti)Risk less part: A deterministic drift.µ = annualized risk less rate of return.
σ ·∆W (ti) = σ · (W (ti+1)−W (ti))︸ ︷︷ ︸Normal distributedwith variance ∆ti
Risky part: A normal distributed randomvariable. σ = annualized standard devi-ation of the return.
© copyright 2012 Christian Fries 43 / 51
![Page 60: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/60.jpg)
Example: Black-Scholes ModelMonte-Carlo Implementation
Implementation: (Monte-Carlo) Simulation: Consider time-discreteversion of Black-Scholes Model
∆S(ti)S
= µ ·∆ti + σ ·∆W (ti)
i.e.S(ti+1) = S(ti) + µ ·S(ti) ·∆ti + σ ·S(ti) ·∆W (ti).
For a single path (scenario) we have
S(ti+1,ω) = S(ti ,ω) + µ ·S(ti ,ω) ·∆ti + σ ·S(ti ,ω) ·∆W (ti ,ω) (Euler Scheme).
Remark: Much more accurate numerical simulation is given by
S(ti+1,ω) = S(ti ,ω) ·exp((µ− 1
2σ
2) ·∆ti + σ ·∆W (ti ,ω))
(Log-Euler Scheme).
© copyright 2012 Christian Fries 44 / 51
![Page 61: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/61.jpg)
Example: Black-Scholes ModelBinomial Tree Implementation
Implementation: (Binomial) Tree: Consider time-discrete version ofBlack-Scholes Model
S(ti+1) = S(ti) + µ ·S(ti) ·∆ti + σ ·S(ti) ·∆W (ti).
Approximate the normal distributed ∆W (ti) by a binomial distributedrandom variable ∆B(ti)
S(ti+1) ≈ S(ti) + µ ·S(ti) ·∆ti + σ ·S(ti) ·∆B(ti).
where
∆B(ti) =
+√
∆ti with probability 12
−√
∆ti with probability 12
Justification: Repetitive binomial experiments converge to a normaldistribution.
n
∑i=0
∆B(ti) ≈n
∑i=0
∆W (ti) for n large.
© copyright 2012 Christian Fries 45 / 51
![Page 62: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/62.jpg)
Example: Black-Scholes ModelUnder Martingale Measure
Back-Scholes Model:
Evolution of Money Market Account: dB = r ·B dt
Evolution of Stock . . . . . . . . : dS = µ ·Sdt + σ ·SdW (t)
Choose Numéraire and consider Martingale Measure:
B chosen as Numéraire ⇒ SB
is Q-martingale
Change of Drift Trick:
dSB
= (µ− r)SB
dt + σSB
dW (t)
drift has to be zeroµ = r
© copyright 2012 Christian Fries 46 / 51
![Page 63: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/63.jpg)
RISK NEUTRAL VALUATIONEXAMPLE: BLACK-SCHOLES MODEL
MONTE-CARLO OPTION PRICER
![Page 64: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/64.jpg)
Example: Black-Scholes ModelExample: Monte-Carlo European Option Pricing
Example: European option value: European option value is a knownfunction of the underlying process(es) at some future time Tn:
V (Tn) = max(S(Tn)−K ,0) .
Choose a Model for Underlyings: e.g. Black-Scholes
Evolution of Money Market Account: dB = r ·B dt
Evolution of Stock . . . . . . . . : dS = µ ·Sdt + σ ·SdW (t)
Choose Numéraire and consider Martingale Measure:
B Numéraire ⇒ SB
Q-martingale ⇒ µ = r
Monte-Carlo Simulation (of the Q dynamics):
B(ti+1,ω) = B(t0) ·exp(r · ti+1)
S(ti+1,ω) = S(ti) + r ·S(ti) + σ ·S(ti) ·∆W (ti ,ω)
for ω = ω1,ω2,ω3,ω4,ω5,ω6, . . . ,ωN (some 1000’s samples).© copyright 2012 Christian Fries 48 / 51
![Page 65: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/65.jpg)
Example: Black-Scholes ModelExample: Monte-Carlo European Option Pricing
Example: European option value: European option value is a knownfunction of the underlying process(es) at some future time Tn:
V (Tn) = max(S(Tn)−K ,0) .
Choose a Model for Underlyings: e.g. Black-Scholes
Evolution of Money Market Account: dB = r ·B dt
Evolution of Stock . . . . . . . . : dS = µ ·Sdt + σ ·SdW (t)
Choose Numéraire and consider Martingale Measure:
B Numéraire ⇒ SB
Q-martingale ⇒ µ = r
Monte-Carlo Simulation (of the Q dynamics):
B(ti+1,ω) = B(t0) ·exp(r · ti+1)
S(ti+1,ω) = S(ti) + r ·S(ti) + σ ·S(ti) ·∆W (ti ,ω)
for ω = ω1,ω2,ω3,ω4,ω5,ω6, . . . ,ωN (some 1000’s samples).© copyright 2012 Christian Fries 48 / 51
![Page 66: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/66.jpg)
Example: Black-Scholes ModelExample: Monte-Carlo European Option Pricing
Example: European option value: European option value is a knownfunction of the underlying process(es) at some future time Tn:
V (Tn) = max(S(Tn)−K ,0) .
Choose a Model for Underlyings: e.g. Black-Scholes
Evolution of Money Market Account: dB = r ·B dt
Evolution of Stock . . . . . . . . : dS = µ ·Sdt + σ ·SdW (t)
Choose Numéraire and consider Martingale Measure:
B Numéraire ⇒ SB
Q-martingale ⇒ µ = r
Monte-Carlo Simulation (of the Q dynamics):
B(ti+1,ω) = B(t0) ·exp(r · ti+1)
S(ti+1,ω) = S(ti) + r ·S(ti) + σ ·S(ti) ·∆W (ti ,ω)
for ω = ω1,ω2,ω3,ω4,ω5,ω6, . . . ,ωN (some 1000’s samples).© copyright 2012 Christian Fries 48 / 51
![Page 67: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/67.jpg)
Example: Black-Scholes ModelExample: Monte-Carlo European Option Pricing
Example: European option value: European option value is a knownfunction of the underlying process(es) at some future time Tn:
V (Tn) = max(S(Tn)−K ,0) .
Choose a Model for Underlyings: e.g. Black-Scholes
Evolution of Money Market Account: dB = r ·B dt
Evolution of Stock . . . . . . . . : dS = µ ·Sdt + σ ·SdW (t)
Choose Numéraire and consider Martingale Measure:
B Numéraire ⇒ SB
Q-martingale ⇒ µ = r
Monte-Carlo Simulation (of the Q dynamics):
B(ti+1,ω) = B(t0) ·exp(r · ti+1)
S(ti+1,ω) = S(ti) + r ·S(ti) + σ ·S(ti) ·∆W (ti ,ω)
for ω = ω1,ω2,ω3,ω4,ω5,ω6, . . . ,ωN (some 1000’s samples).© copyright 2012 Christian Fries 48 / 51
![Page 68: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/68.jpg)
Example: Black-Scholes ModelExample: Monte-Carlo European Option Pricing
Calculate Payoff for each Sample Path:
V (Tn,ωj) = max(S(Tn,ωj)−K ,0
)Calculate Price:
V (0) = B(0) ·EQ(
V (Tn)
B(Tn)
)≈ 1
N
N
∑j=1
V (Tn,ωj)
B(Tn,ωj)
=1N
N
∑j=1
max(S(Tn,ωj)−K ,0
)B(Tn,ωj)
© copyright 2012 Christian Fries 49 / 51
![Page 69: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/69.jpg)
Example: Black-Scholes ModelExample: Monte-Carlo European Option Pricing
Exercise: Excel Sheet: Monte-Carlo Simulation with Option Pricing
© copyright 2012 Christian Fries 50 / 51
![Page 70: Risk Neutral Valuation - Christian Frieschristian-fries.de/finmath/lecture12/Fries - Presentation... · 2019. 12. 15. · Christian Fries](https://reader034.vdocument.in/reader034/viewer/2022051922/600fdff021e5226de9320774/html5/thumbnails/70.jpg)
Further Reading I
FRIES, CHRISTIAN P.: Mathematical Finance. Theory, Modeling,Implementation. John Wiley & Sons, 2007. ISBN 0-470-04722-4.http://www.christian-fries.de/finmath/book.
© copyright 2012 Christian Fries 51 / 51