-----------------------------------------------------
# #
# Chapter 8: Stochastic Calculus #
# and the Ito-Formula #
# #
-----------------------------------------------------
Summary: Recall the main result of the first chapter
on trading strategies: Hold delta_k stocks at the end
of day t_k and close the position on day t_N. Suppose
V_0 is your initial money. Then at time t_N this stra-
tegy has generated the (positive or negative) amount
V_N = V_0 + sum_{k=1}^N delta_{k-1}*(S_k - S_{k-1}) (1)
if S_k denotes the closing price of the traded stock
on day t_k. Here we assumed zero interest rates. For
non-zero interest rates, the quantities V and S in (1)
above have to be substituted by discounted quantities.
Now assume that the underlying price is modelled by a
Black-Scholes model in continuous time. Then the sum
on the right hand side of (1) is a sum over stochastic
quantities. If we consider the continuous time limit
of (1), then an obvious notation for (1) would be
V_T = V_0 + int_0^T delta_t * dS_t (2)
with a stochastic differential dS_t. Apparently, in
equation (1) it makes a difference whether we have
delta_{k-1} = delta_{k-1}(s_0,...,S_{k-1}) (3a)
meaning we can use past or current information, or
delta_{k-1} = delta_{k-1}(s_0,...,S_{k-1},S_k) (3b)
which would mean that we can use future information:
The case (3b) would mean that when we adjust our stock
position at the end of day t_{k-1}, we already know
which underlying price S_k will realize tomorrow and
of course this information then can be used to make
unlimited profit. Thus, mathematically, it is a crucial
requirement that the delta's showing up in (1) or then
in the continuous time version (2), fulfill the require-
ment (3a), they cannot depend on future information.
In the more mathematical framework of stochastic inte-
grals, the condition (3a) leads to the notion of Ito-
integrals, which are, at least in the context of option
pricing, apparently the right ones to work with, whereas
a more general condition (3b) may lead to so called
Stratonovich integrals, and, in the continuous time limit,
these objects have indeed different limits. From a prac-
tical point of view this sounds quite obvious, but from
a more mathematical perspective, for someone who is used
to numerically calculate standard (that is, non-stocha-
stic) integrals just by discretizing to Riemannian sums,
this sounds a bit surprising. Thus, in this chapter
these issues are considered more closely.
pdf-file: Chapter 8: Stochastic Calculus and the Ito-Formula