Validating a Bootstrapped Discount Curve

Summary

This note offers two practical checks for a bootstrapped interest-rate discount curve built from market instruments such as cash deposits, futures, bonds, or swaps. First, reprice the instruments used as inputs and confirm that the curve reproduces their market quotes. This tests whether the bootstrap has fitted its calibration instruments.

Next, inspect implied forward rates, for example by plotting daily three- or six-month forwards and looking for irregular behavior around curve nodes. The note cautions that forward shapes depend on the interpolation method and points to specialist references on yield-curve construction and multiple-curve bootstrapping. These checks are useful diagnostics, but the brief answer does not define quantitative acceptance thresholds or establish that smooth forwards alone make a curve fit for every purpose. Validation should therefore be understood as calibration and shape review, with suitability depending on the curve’s intended use.

Key ideas

• Reprice the calibration instruments to check that the curve reproduces their market quotes.
• Inspect forward rates around curve nodes for irregular behavior.
• Plotting daily forwards can help reveal interpolation artifacts.
• Forward-rate shapes depend on the interpolation method used.
• The note offers qualitative checks but no numerical acceptance thresholds.

Tags

# How to Validate and Test a Discount curve (i.e. SOFR, LIBOR, ESTR)


# How to Validate and Test a Discount curve (i.e. SOFR, LIBOR, ESTR)












Let's say an Interest Rate/Discount Curve (SOFR, ESTR, LIBOR or any other) is bootstrapped using the standard inputs and market quotes for Cash, Futures, Bonds/Swaps. What are the metrics to be looked at, steps to be followed and tests to be done in order to validate that the bootstrapped curve is valid and fit for purpose? (any literature or practitioner references would be highly appreciated).

## Answer by David Duarte (score 2)

https://quant.stackexchange.com/a/63646

A first step would obvisouly be to check if the curve you built replicates the input instruments.

A second step might be to check the forwards to see if there is irregular behaviour around the curve nodes. Look at a plot of daily 3m or 6m forwards which should be smooth. Different interpolation methods will generate diferent results and for this might I suggest "Methods for Constructing a Yield Curve" by Patrick S. Hagan and Graeme West.

Another good reference on curve construction would be "Everything You Always Wanted to Know About Multiple Interest Rate Curve Bootstrapping but Were Afraid to Ask" by Ametrano and Bianchetti

Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)

This summary was written by Stratmill's research agent from the original; it is not a copy of the source.