Hello all ... I am attempting to price an option on 10yr CMS via Monte Carlo simulation using a 2F HW model.The option pays off 100bps at the end of each year for the next 10 years if 10yr CMS at that time is < 4.50%. The approach I was taking is as follows:1. i = 1 to N2. j = 1 to 103. Draw random variates, simulate evolution of yield curve from year j-1 to year j4. Analytically calculate 10yr swap rate from yield curve5. Calculate payoff6. Next j7. Sum discounted payoffs for j = 1 to 10, store as f(i)8. Next iThe average of my f(i)'s should be the option value. Problem is, I am not close to the observed price.Any issues with my methodology?Rgds

cos ur not pricing in the skew effectwhat are you calibrating to

Thanks inarrears for your reply.You are correct, I was calibrating to ATMFs. Obviously there are a ton of other things I could be calibrating to ... what do you recommend?Rgds

If I well understand, it' s a sum of digitals :at each time i you receive 1% if the CMS at this time is < 4.50% , 0 if not ???In this case, write (CMS - K, 0 ) = f(K, sigma(K)) then 1{CMS>K} = - df/dK - (df/dsigma)*(dsigma/dK)so your payoff is a call on spread plus a smile component...You didn't need to run a long monte carlo...and you can easy capture the smile effect...Kipi...

Thanks Kipi for your reply ... actually for this purpose you are correct ... but I was building the model more generally to account for the structures which are not digital but pay off for example MAX(CMS30 - CMS10, 0).Do you have any insights on the skew, etc .. ???Rgds

According to me, the best way to price this product is to use a stochastic vol model who can handdle the smile....With a simpler model, i think you can't do this: Your price will be far away...Once your stochastic model is calibrated to the swaption smile, you just run a Monte carlo to obtain your price...But it's quite difficult to build a stochastic model, properly calibrate to the whole swaption smile. For the moment, i don't achieve to do this correctly.Vlad piterbarg has build a very strong stochastic LMM model that allow to capture the whole swation smile, it's seems to be promising....Kipi...