MATLAB: Does the Nelson–Siegel–Svensson model give me different parameters on different computers for the same data

fitsvenssonMATLAB

I am trying to fit the Nelson-Siegel-Svensson model to my data using the "fitSvensson" command, but I get different fitting parameters when I run the same code on different computers.

Why are these results not consistent and are there any steps I can take to ensure that I always get the same fitting parameters for my data?

Best Answer

Nelson-Siegel-Svensson model is particularly hard to fit to data, as multiple combinations of fitting parameters can result in models with a similar 'goodness of fit' to the data (see Gilli et al. "Calibrating the Nelson-Siegel-Svensson Model").

Fitting the Svensson model to data if fundamentally a nonlinear optimization problem and in this particular case very minor variations in how the optimal parameters are calculated can result in very different local optima being found. These variations can be as minor as differences in how computers store numbers in their memory.

The best way to limit variation in fitting parameters using the "fitSvensson" function is to specify tight bounds on the acceptable fitting parameters. This will force the solver to converge to a specific local optimum.

Related Solutions

MATLAB: What is the difference between IMPVBYBLS and BLSIMPV, and how to I determine the Black-Scholes implied volatility using these functions

Both IMPVBYBLS and BLSIMPV use the Black-Scholes model to compute the implied volatility for a given option price. However, one of the main differences between the two functions is that IMPVBYBLS allows specifying cash dividends as well as continuous dividend yields, while BLSIMPV only allows continuous dividend yields.

If you specify continuous dividend yields for both IMPVBYBLS and BLSIMPV, they produce the same results:

Strike = 29;
AssetPrice = 30;
Sigma = .25;
Rates = 0.05;
Settle = datenum('01-Jan-2008');
Maturity = datenum('01-May-2008');
RateSpec = intenvset('ValuationDate', Settle, 'StartDates', Settle, ...
'EndDates',Maturity, 'Rates', Rates, 'Compounding', -1,'Basis',1);
DividendType = {'continuous';'continuous'};
DividendAmounts = [0.05; 0.045];
ExDividendDates = {NaN;NaN};
StockSpec = stockspec(Sigma, AssetPrice, DividendType, DividendAmounts);
OptSpec = {'call'; 'put'};
OptionPrice = optstockbybls(RateSpec, StockSpec, Settle, Maturity, OptSpec, Strike);
ImpvVol1 = impvbybls(RateSpec, StockSpec, Settle, Maturity, OptSpec, ...
Strike, OptionPrice)
ImpvVol2 = blsimpv(AssetPrice, Strike, Rates, ...
yearfrac(Settle,Maturity,1), OptionPrice, 'Class', OptSpec,'Yield',DividendAmounts)
ImpvVol1 =
0.250000000000000
0.250000000000000
ImpvVol2 =
0.250000000000000
0.250000000000000

MATLAB: Is it possible to fit an interest rate curve using the Nelson-Siegel or Svensson parametric models in MATLAB

It is possible to fit an interest rate curve with the Nelson-Siegel or Svensson parametric models using the FITNELSONSIEGEL and FITSVENSSON methods of the IRFUNCTIONCURVE class in the Fixed-Income Toolbox V1.7 (R2009a).

For an example on how to use these methods, please refer to the MATLAB documentation page for each of these functions by typing the follownig command on the MATLAB Command Prompt:

doc fitNelsonSiegel
% or
doc fitSvensson

Best Answer

Related Solutions

MATLAB: What is the difference between IMPVBYBLS and BLSIMPV, and how to I determine the Black-Scholes implied volatility using these functions

MATLAB: Is it possible to fit an interest rate curve using the Nelson-Siegel or Svensson parametric models in MATLAB

Related Question