The optimal-drift model: an accelerated binomial scheme

We introduce the optimal-drift model for the approximation of a lognormal stock price process by an accelerated binomial scheme. This model converges with order o(1/N), which is superior compared to today’s benchmark methods. Our approach is based on the observation that risk-neutral binomial schemes converge to the lognormal limit independently of the choice of the drift parameter. We verify the improved order of convergence by an asymptotic expansion of the binomial distribution function. Further, we show that the above result on drift invariance implies weak convergence of the binomial schemes suggested by Tian (in J. Futures Mark. 19, 817–843, <CitationRef CitationID="CR11">1999</CitationRef>) and Chang and Palmer (in Finance Stoch. 11, 91–105, <CitationRef CitationID="CR4">2007</CitationRef>). Copyright Springer-Verlag 2013