Methodology
Data Collection: Taken from Saudi Exchange (Tadawul) government bond and sukuk transactions at 5:00 PM KSA.
Data Filtering: Only confirmed trades are considered for bootstrapping and curve-fitting.
Historical Aggregation: When fewer than 3 bonds trade on a given day, we aggregate trades from up to 7 prior days (unique ISINs only) to ensure adequate data points for curve fitting.
Curve Fitting: Nelson-Siegel parametric model with volume-weighted optimization. Bonds are weighted by relative trading volume; SAMA reverse repo rate provides a short-end anchor.
Temporal Smoothing: Alpha (confidence score) ranges from 0.0 to 1.0 based on: bond count (20%), volume (50%), maturity coverage (15%), fit quality (10%), and same-day trading proportion (5%). Smoothed curve = α × today’s curve + (1-α) × previous curve.
Full technical documentation →
Appropriate Use
Follow judgement:
- IAS 19 discount rate determination for employee benefit obligations
- IFRS compliance and financial reporting
- Actuarial valuations requiring SAR-denominated discount rates
- Benchmarking and economic analysis
Not Appropriate For
Any other purpose not listed above, including:
- Trading, investment decisions, or portfolio management
- Pricing derivatives or structured products
- High-frequency or intraday applications
- Regulatory capital calculations requiring approved vendor data
Disclaimers
Intended users: This data is provided exclusively for Numerica clients and their professional advisors working on IFRS/IAS 19 compliance matters.
No warranty: No representations or warranties regarding accuracy, completeness, or timeliness. Yield curves are fitted models; thin trading may affect reliability.
Professional judgment required: Users must exercise independent professional judgment and verify data before use in financial reporting. Ultimate responsibility for discount rate selection rests with the reporting entity and their auditors.
No liability: Numerica Consulting shall not be liable for losses, damages, or adverse consequences arising from use of this data.