| Introduction | Uniform Distribution | Classical Examples | Construction | Irreducible Polynomials | Numerical Integraion | Pricing Theory | References |
The risk-neutral valuation paradigm shows that the fair value at time
of a financial derivative maturing at time
is expressed as an integral. For example, for a European option with
pay-off function
on an underlying with initial price
and volatility
where
is the riskless interest rate over the period
, the function
the density
and
is the inverse cummulative normal distribution function (entering from
the assumption that the price process
of the underlying satisfies a geometric Brownian motion).
![[Graphics:../Images/QRDemonstration_gr_382.gif]](../Images/QRDemonstration_gr_382.gif)
In case of a European call option with strike
the function
(depending on
,
,
,
, and
) becomes
![[Graphics:../Images/QRDemonstration_gr_390.gif]](../Images/QRDemonstration_gr_390.gif)
Let us extract a one-dimensional sequence of quasi-random numbers from the previously calculated Niederreiter sequence in base 3 and calculate the sample mean:
![[Graphics:../Images/QRDemonstration_gr_391.gif]](../Images/QRDemonstration_gr_391.gif)
![[Graphics:../Images/QRDemonstration_gr_392.gif]](../Images/QRDemonstration_gr_392.gif)
If we compare this to the analytic solution obtained from the Black-Scholes formula we see a high accuracy of quasi-Monte Carlo integration:
![[Graphics:../Images/QRDemonstration_gr_394.gif]](../Images/QRDemonstration_gr_394.gif)
![[Graphics:../Images/QRDemonstration_gr_396.gif]](../Images/QRDemonstration_gr_396.gif)
![[Graphics:../Images/QRDemonstration_gr_397.gif]](../Images/QRDemonstration_gr_397.gif)
![[Graphics:../Images/QRDemonstration_gr_399.gif]](../Images/QRDemonstration_gr_399.gif)
We visualize how good the first
(transformed) sequence elements of a Niederreiter sequence can
approximate the Gauss distribution. To this end, we throw away the first sequence element
(as it is 0) and invert with the inverse cummulative distribution function.
![[Graphics:../Images/QRDemonstration_gr_402.gif]](../Images/QRDemonstration_gr_402.gif)
![[Graphics:../Images/QRDemonstration_gr_403.gif]](../Images/QRDemonstration_gr_403.gif)
![[Graphics:../Images/QRDemonstration_gr_405.gif]](../Images/QRDemonstration_gr_405.gif)
bars = BarChart[buckets,PlotRange->All,BarLabels->None,BarEdges->False];
![[Graphics:../Images/QRDemonstration_gr_406.gif]](../Images/QRDemonstration_gr_406.gif)
Finally we superimpose the Gaussian density function with mean and variance the sample
mean respectively the sample variance of the
transformed samples.
![[Graphics:../Images/QRDemonstration_gr_408.gif]](../Images/QRDemonstration_gr_408.gif)
![[Graphics:../Images/QRDemonstration_gr_409.gif]](../Images/QRDemonstration_gr_409.gif)
Converted by Mathematica November 18, 1998
| Introduction | Uniform Distribution | Classical Examples | Construction | Irreducible Polynomials | Numerical Integraion | Pricing Theory | References |