Dear All,I am looking for help with the following:I wrote a simple function that discounts all the flows from a given bond using just a bond's payment schedule and 6 parameters of zero coupon curve (nelson siegel form - with a humb). Now I usually "fit" the function's results of series of bonds with empirical values from the market by changing the parameters of NS zero curve to minimize the differences between the model and market. It also takes 20-30 seconds to produce best fit answers. I know that using SOLVER for such a calculation is a very primitive and inaccurate solution. Does anybody have any expirience with solving that problem in a clever way using matrices or some smart add-in ?Thank you in advanceMD

Hi mdecSolver have many problems. Use a factor ( 100000) multiplying the errors.some time this increase the accuracy of the solution.www.solver.com Take a look to this add-in