wilmott.com

i have spot rates derived from the nelson and siegel approach:s = b0 + b1*(1 - exp(-t))./(t) + b2*((1 - exp(-t)) ./(t) - exp(-t)) ;for simplicity assume the bond is maturing after 0.8 yearsThe question is if I want to price the semi annual coupon bond with these spot rates. what would the price be? I computed the spot rates s1 and s2 by plugging b0,b1,b2 of nelson and siegel and 0.3 and 0.8 for time respectively.the price of the bond is c*d1 +(c+100)*d2 where c is the coupon payment (i.e coupon rate/2) and d1 and d2 are the discount rates of periods 0.3 and 0.8 respectivelythe question is :to compute d1 and d2 should I use the formula:d1 = exp( -(s1).*0.3 )and d2=exp( -(s2).*0.8 )?is that correct?

if you have spot rate derived from the curve then you must know what the basis of the rates are ? How did you derive them ? I have never used Nelson-Siegel but I would assume that you need to do some form LS fitting.

the NS parameters were given to me. they give spot rates with continuous compounding. But i am wondering if I can price a bond using them as it was described in my previous thread

if they are continuously compounded then yes you can use them as you have specified. you might need to be a bit careful with the time values.. they might not be 0.3 and 0.5 of a year but you should be OK

I am a bit worried by exp(-S/2)*t for discount rates gives an answer close to the observed one whereas the exp(-s)*t discount is giving me an answer far from the observed price. Can you double check whether the spot should be divided by two or not when I want to compute the bond price?

you seem to have an error in your formula - on Wikipedia this is what Nelson and Siegel looks likewhereas in your formula the last exp is outside the brackets

I don't see the difference between the two . m/tau is my t and the exponential is inside the bracket (exp(-t) is multiplied by b2)

your are right my bad...