Skip to content
All library documents

Why the G2++ Short-Rate Model Misses Volatility Smile and Skew

Article Quant Q&A · Author: FunnyBuzer

Summary

The document introduces the two-factor G2++ short-rate model, in which the short rate combines two mean-reverting Gaussian factors with a time-dependent shift. It notes that the model can reproduce several stylized features of the yield curve, yet may not represent the volatility smile or skew well. The central question is whether extensions can improve smile and skew fit while retaining the G2++ framework.

No extension, calibration procedure, or empirical comparison is supplied; the text poses the modeling question rather than answering it. It names Cheyette, SABR, and ZABR as alternatives the questioner would prefer not to adopt, but gives no evidence that any particular modification works. Readers should treat this as a prompt for further model research, not as a proposed strategy or a demonstrated result.

Key ideas

  • G2++ represents the short rate as two mean-reverting factors plus a deterministic shift.
  • The question raises a perceived limitation in fitting rate-option volatility smile and skew.
  • It asks about extensions that preserve the G2++ model family.
  • The document proposes no extension or supporting evidence.

Tags

Full text
# Produce volatility smile/skew with G2++ model


# Produce volatility smile/skew with G2++ model












Suppose I have a G2++ short rate model: $$r(t)=x(t)+y(t)+\phi(t), \quad r(0)=r_0$$ with $$dx(t)=-ax(t)dt+\sigma dW_1(t), \quad x(0)=0$$ $$dy(t)=-bx(t)dt+\eta dW_2(t), \quad y(0)=0$$ $$d\langle W_1,W_2\rangle_t = \rho dt$$ Although it has many features that are typical stylized facts of the yield curve, the does not seem to capture well the volatility smile/skew. I would like to know whether there are some model extensions enabling to partly capture the smile/skew, without switching to a different model, such as Cheyette, SABR or ZABR.

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.