February 3rd, 2011, 1:15 am
I haven't tried it. But, the volatility x(t) = sig(t) or the variance rate x(t) = sig(t)^2 follows an sded x = x ( a - b log x) dt + c x dW(t), with (a,b,c) some constants.so your expansion parameter will be c. From dimensional analysis, you probably want c^2 T < 1.You can also use dimensional analysis to estimate c for your application if youknow the corresponding vol. of vol. for, say the Heston model.(See the discussion in my book bottom of pg 4).