Serving the Quantitative Finance Community

 
User avatar
brianhclo
Topic Author
Posts: 11
Joined: March 28th, 2006, 11:25 pm

Reverse Monte Carlo

September 19th, 2006, 11:15 am

I am trying to price a structure using monte carlo. the idea is that we input the coupon and simulate the fair value.Now, I changed the logic of the payoff where we now input the FV and simulate the coupon. (everything staying the same)However my results are not matching. ie by subbing the Coupon back into the original spreadsheet I am not getting the same FV.Can anyone offer any advice?Thanks
 
User avatar
Panang
Posts: 150
Joined: June 8th, 2006, 11:22 am

Reverse Monte Carlo

September 19th, 2006, 1:37 pm

Are you using the same set of random numbers?
 
User avatar
brianhclo
Topic Author
Posts: 11
Joined: March 28th, 2006, 11:25 pm

Reverse Monte Carlo

September 19th, 2006, 3:22 pm

the random numbers should not matter as i am running 25000 paths.i suspect it is the standard deviation which causes this error. this is because for a FV of 100 the 2 methods match but the further we go away from 100 the higher the standard deviation and the bigger the error between 2 methods.has anyone come across this standard deviation problem before?thanks,brian
 
User avatar
Traden4Alpha
Posts: 23951
Joined: September 20th, 2002, 8:30 pm

Reverse Monte Carlo

September 19th, 2006, 8:20 pm

1) Any chance that when you reverse the Monte Carlo, you are subtracting the market movements in a way that isn't an exact reverse of the forward simulation? 2) Could the simulation be creating a skewed distribution of outcomes so that when you substitute a fixed FV for the forward sim's FV distribution, the reverse sim skews the data in the other way and biases the coupons?3) You mention that the forward = the reverse for FV=100 and departs otherwise. If so are you normalizing some part of the calculation but not normalizing all the facets of the sim that need to be normalized?