Spot term structure is rarely available as its methodological procedures are complicated and therefore market participants substitute it with traditional term structure as an approximation. This paper proposes two hybrid optimisation algorithms consisting of traditional algorithms to reduce complex procedures. Estimated by any algorithm, curve model parameters are always prone to overshooting risks. We modify the algorithms to overcome the risks and use the algorithms to estimate on-the-run government bond data of Indonesian domestics, Yankees as well as the US Treasuries from 17th April 2013 to 29th October 2013. To confirm the validity of term structure measurement, we propose to standardize three Bolder and Stréliski criteria as there is no agreement for performance criteria. We modify goodness of fit and robustness indicators of the criteria to do reliable assessments. We have implemented the modified Bolder and Stréliski criteria to compare the performances of the algorithms and to distinguish liquid market data, the US Treasuries, from less liquid markets, Indonesian government bonds.
- Hybrid optimisation algorithms
- Modified Bolder and Stréliski performance indicators
- Monte Carlo method
- Nelson Siegel extensions
- Newton method
- Term structure of interest rates