Hi;I wonder that Can we use different Probability distributions in GBM for dW term?Just only using for Normal Distribution or using another distribuiton for Currency simulation?Thanks all..

Sure -- just don't call them GBM. Here are two approaches.I. dW draws are independent and identically distributed (iid), right? A general class of iid processes (in continuous time) are called Levy processes, written say L(t). So, roughly dW -> dL is what you want.Google "exponential Levy models"; there is also a Levy Process thread in the FAQ forum.Simulating them is discussed in Schouten's 'Levy Processes in Finance' and Cont/Tankov's 'Financial Modelling with Jump Processes'II.Alternatively, suppose you don't require a continuous-time limit, but have a fixed time step Dt .Then, you can draw from whatever (zero-mean) distribution you want, written say F(x) for the distribution of X = log S(t)/S(t-Dt).For example, maybe you will parameterize and risk-adjust the historical distribution for time step Dt.This is just an iid random walk model.So, now dW -> dF is what you want. Simulation can be done by the Inverse methodregards,

Thanks Alan;First of all dW draws independent and identically distributed I am ok.In continuous time process simulation any process you mean that Jump process has to be transform for other distribution inverse(iid),ok.S*Exp((mu-sigma^2)*t+vol*sqr(t)*normınv(0,1)-------I say this distribution can be change for any continuous time process.Finally If I inverse transform I can use distribution like(logistic distribution,student-t distribution etc.),can't I?Thanks.