Nelson Siegel Svansson
📚 Nelson-Siegel-Svannson algorithm 📚
Popular algorithm for fitting a yield curve to obseved data.
Problem
Data on bond yields is usualy avalible only for a small set of maturities, while the user is normaly interested in a wider range of yields.
Solution
A popular solution is to use an algorithm to find a function that fits the existing datapoints. This way, the function can be used to interpolate/extrapolate any other point. The Nelson-Siegel-Svannson model is a curve-fitting-algorithm that is flexible enough to approximate most real world applications.
The Nelson-Siegel-Svensson is an extension of the 4-parameter Nelson-Siegel method to 6 parameters. The Scennson introduced two extra parameters to better fit the variety of shapes of either the instantaneous forward rate or yield curves that are observed in practice.
Advantages:
- It produces a smooth and well behaved forward rate curve.
- The intuitive explanation of the parameters.
beta0
is the long term interest rate andbeta0+beta1
is the instantaneous short-term rate.
To find the optimal value of the parameters, the Nelder-Mead simplex algorithm is used (Already implemented in Matlab's fminsearch function). The link to the optimization algorithm is Gao, F. and Han, L. Implementing the Nelder-Mead simplex algorithm with adaptive parameters. 2012. Computational Optimization and Applications. 51:1, pp. 259-277.
The furmula for the yield curve (Value of the yield for a maturity at time 't') is given by the formula:
Parameters
- Observed yield rates
YieldVec
. - Maturity of each observed yield
TimeVec
. - Initial guess for parameters
beta0
,beta1
,beta2
,beta3
,labda0
, andlambda1
. - Target maturities
TimeResultVec
.
Desired output
- Calculated yield rates for maturities of interest
TimeResultVec
.
Getting started
The user is interested in the projected yield for government bonds with a maturity in 1,2,5,10,25,30, and 31 years. They have data on government bonds maturing in 1,2,5,10, and 25 years. The calculated yield for those bonds are 0.39%, 0.61%, 1.66%, 2.58%, and 3.32%.
TimeVec = [1; 2; 5; 10; 25];
YieldVec = [0.0039; 0.0061; 0.0166; 0.0258; 0.0332];
beta0 = 0.1; % initial guess
beta1 = 0.1; % initial guess
beta2 = 0.1; % initial guess
beta3 = 0.1; % initial guess
lambda0 = 1; % initial guess
lambda1 = 1; % initial guess
TimeResultVec = [1; 2; 5; 10; 25; 30; 31]; % Maturities for yields that we are interested in
%% Implementation
OptiParam = NSSMinimize(beta0, beta1, beta2, beta3, lambda0, lambda1, TimeVec, YieldVec); % The Nelder-Mead simplex algorithem is used to find the parameters that result in a curve with the minimum residuals compared to the market data.
disp("Optimal parameters are")
disp(OptiParam)
ResultYield = NelsonSiegelSvansson(TimeResultVec, OptiParam(1), OptiParam(2), OptiParam(3), OptiParam(4), OptiParam(5), OptiParam(6)); % Calculate the yield for tergeted maturities using the calibrated partameters
Cite As
Gregor Fabjan (2024). Nelson Siegel Svansson (https://github.com/qnity/nelson_siegel_svansson_matlab), GitHub. Retrieved .
MATLAB Release Compatibility
Platform Compatibility
Windows macOS LinuxTags
Community Treasure Hunt
Find the treasures in MATLAB Central and discover how the community can help you!
Start Hunting!Discover Live Editor
Create scripts with code, output, and formatted text in a single executable document.
Versions that use the GitHub default branch cannot be downloaded
Version | Published | Release Notes | |
---|---|---|---|
1.0.1 | Project now has a website |
|
|
1.0.0 |
|