An approximate approach to fractional analysis for finance (Q2490081)

The author studies the fractional process \(B_t=\int_0^t(t-s)^{H-1/2}dW_s\) where \(W\) is standard Brownian motion and \(H\), \(0<H<1\), is the Hurst index. He proposes the approximation \(B_t^\varepsilon=\int_0^t(t-s+\varepsilon)^{H-1/2}dW_s\) (\(H\neq \frac{1}{2}\)), for every \(\varepsilon > 0\), of \(B_t\) and shows that the process \((B_t^\varepsilon,\,t\geq 0)\) is a semimartingale, in contrast to the process \((B_t,\,t\geq 0)\), and that it converges to \(B_t\) in \(L^2(\Omega)\) for \(\varepsilon \rightarrow 0\). Next the fractional Langevin equation driven by \(B_t\) and an approximated version driven by \(B_t^\epsilon\) are discussed. Further the author replaces the Wiener process in the standard model of stock price dynamics \(dS_t= S_t(\mu dt + \nu dW_t)\) by \(B_t\) and derives an approximate model using \(B_t^\varepsilon\) and the explicit solution of the latter. Finally he discusses the possibility of arbitrage in these models, the fractional one driven by \(B_t\) and the approximate one driven by the semimartingale \(B_t^\varepsilon\).