Hi,I hope my question doesnt seem too silly, but it seems like Hull white's vol is always calibrated to european options. May I know why this is so? Why doesnt pple calibrate it to american options etc?

Ok. I think I have found the answer myself. Coz hull white is not used to calculate american options.On another related qn.... if we calibrate hull white to purely european options, how accurate is it to use hull white to compute the price of a bermudan? By using hull white with vols from european options to compute a bermudan, arent we sort of making an assumption that the vol of a bermudan is the same as that of a european?

I'm working in solving a range accrual callable. In fact I want to put a price to this kind of option. I've thought use a Hull White's model but I've some doubts. I use this model:dr(t)=a(R(t)-r(t))dt+Sdz(t)I know I have to callibrate a and S but I don't really is R(t). I have read that this parameter has to be choosen to adjust the zero coupon bonds and in other papers I have found the next expression: R(t)=f(t0,t)+(1/a)*(df(t0,t)/dt)+(S^2/2a^2)*(1-exp(-2a*(t-t0))Somebody know what exactly is this parametre?I'm great of some answer. Thanks million

Late answer but anyway:you found the right answer, the parameter R(t) is used to recover exactly the initial yield curve.Indeed the sde dr(t) gives you r(t) in terms of R(t), and then you can calculate the zero coupon bond bythe affine formula P(t, T) = exp( A(t, T) - B(t, T) r(t) ),with A(t, T) and B(t, T) having a definite expression in terms of the parameters. Therefore the bond pricewill depend on your choice of R(t) through r(t). If you take R(t) = b where b is a constant then you get Vasicek model,which cannot accomodate any curve, and actually makes predictions on the form of the initial curve.However if you assume that P(0,T) is a known function, then you can recover the expression that you found in the literature for R(t) by forcing the equality of your known P(0,T) with its affineexpression above.