Skip to content
All library documents

Choosing Conventional CDS Spreads for Sentiment Regression

Article Quant Q&A · Author: mkoall
Read the original

Summary

The document considers how to relate sentiment derived from text analysis to a firm's credit default swap spreads. The question is whether to use par spreads or conventional spreads in a regression, and it notes that both may be quoted in basis points or decimal form.

The answer recommends conventional spreads, describing par spreads as a legacy measure associated with the period before CDS market standardization. It also points to a related study of executive remarks and CDS spreads. The discussion is brief: it does not define either spread, explain their construction or units in detail, or provide regression results. Researchers should verify the appropriate series and quote convention for their data source and sample before using the recommendation.

Key ideas

  • The proposed analysis relates text-based sentiment to firm-level CDS spreads.
  • The answer recommends conventional spreads rather than par spreads for the regression.
  • The document characterizes par spreads as a legacy measure from before CDS standardization.
  • It does not explain spread construction or demonstrate how the choice affects regression results.

Tags

Fixed incomeDerivatives pricingSentiment
Full text
# Markit CDS price data


# Markit CDS price data












I am currently looking at a way to run a basic sentiment analysis NLP model and look for a possible relationship with firm CDS spreads. I am however unsure as to which spread to use for my regression (par-spread,convspreard). Especially since both parspread and convspread seem to be quoted as follows i.e 12bps =0.00012

Thank you in advance

## Answer by Dimitri Vulis (score 2, accepted)

https://quant.stackexchange.com/a/75074

Related recent paper: https://www.spglobal.com/marketintelligence/en/news-insights/research/watch-your-language-executives-remarks-on-earnings-calls-impact-cds-spreads

Use conventional spreads. Par spreads are legacy, going back to before the Big Bang. No one should be using them.

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.