Fractional Brownian motion as a weak limit of Poisson shot noise processes -- with applications to finance (Q2485795): Difference between revisions

The authors consider weak convergence of some Poisson shot noise processes \((S_xx(t))_{t\geq 0}\) that are appropriate to model stock prices and provide an economic reason for long-range dependence in asset returns. More specifically, let \(g:\mathbb{R}_+\to\mathbb{R}\) be a continuously differentiable function with \(g'(u)= O(u^{-(1/2)-\varepsilon)}\) \((\varepsilon> 0)\). Define stochastic processes \((X_i)= (X_i(t))_{t\geq 0}\) by \(X_i(u)= g(u)Y_i\) \((u\geq 0)\), \(X_i(u)= 0\) \((u< 0),\) where the \(Y_i\) are i.i.d. innovations with \(EY_1= 0\) and \(EY_1^2\in(0,\infty)\). Let \(N\) be a two-sided homogeneous Poisson process with rate \(\alpha> 0\) and \(\cdots< T_{-2}< T_{-1}< 0< T_1< T_2\cdots\) be random points. We assume that the \(X_i\) are independent of \(N\) and \(\{T_i\}\). Consider a shot noise model \((S(t))_{t\geq0}\) defined by \[ S(t)= \sum^{N(t)}_{i=1} X_i(t- T_i)+ \sum^{-\infty}_{i=-1} \{X_i(t- T_i)- X_i(- T_i)\}\quad (t\geq 0),\quad S(0)= 0 \] and put \(S_x(t)= S(xt)/\sqrt{\text{Var}(S(t))}\). The authors prove that \(S(t)@>d>> B^H\) \((t\to\infty)\) in \(D[0,\infty),\) where \(B^H\) is a fractional Brownian motion with Hurst parameter \(H=\gamma+ (1/2)\) \((0< \gamma< 1/2)\).