February 7th, 2011, 4:48 pm
I have heard about a method using cubic spline. Is it closer to what you are searching ? Quite hard to implement it seems.A very easy way could be to have a smoothed volatility smile (with the SVI method of Gatheral for example), to interpolate the volatity smile with the fitted formula for a high resolution curve . Then using BS formula, you transform your vol smile into a price curve and then you numerically twice differentiate that to obtain the implied PDF. Sounds naive but in high resolution it works not so bad it seems.
Last edited by
frenchX on February 6th, 2011, 11:00 pm, edited 1 time in total.