I'm having some trouble pricing a 0% interest rate Floor following Black's formula. The term d1 contains the expresion Ln(Forward/Strike) if the strike is exactly 0 this expresion yields an indetermination and therefore we can't compute N(d1) in the pricing formula.

I was wondering how to work around this. A peer suggested to compute the regular formula using a 100% probability, but the results are not really meningful.

Aside from the displaced model is there any other adjustment to the black model to price a 0% interest rate floor?

  • $\begingroup$ Usually for Floors, strike is given in absolute terms with reference to the at-the-money (ATM) level. In your case, the strike would therefore be Forward + 0% i.e. your floor is ATM and Ln(Forward/Strike) = 0. $\endgroup$ – JejeBelfort Nov 6 '17 at 13:21
  • $\begingroup$ Hi, in my case I only have 0% interest as a given Floor. From what you said I understand that we have 100% probability to execute the Floor which will translate in a N(d2) = 1 and then plug the 1 in BS formula. Is that what you mean? $\endgroup$ – Ladislao Vidal Nov 6 '17 at 14:25
  • $\begingroup$ Forget about this 100% probability, just plug in Black's formula the relevant parameters (i.e. Forward, Strike, Implied volatility and time-to-maturity) and compute the floor price. $\endgroup$ – JejeBelfort Nov 6 '17 at 14:47
  • $\begingroup$ Still not sure about how to just "plug" the Strike since this is 0% d1 is simply undetermined $\endgroup$ – Ladislao Vidal Nov 6 '17 at 15:54
  • $\begingroup$ No, the strike is equal to the forward price as you have an absolute strike of 0% $\endgroup$ – JejeBelfort Nov 6 '17 at 16:25

If the forward is $> 0$ then under the Black & Scholes model the probability of underlying rate being $\leq 0$ is zero, so that the $0\%$ strike floor is worth zero. If the forward is $\leq 0$ (as has been happening since rates went into the negative territory) then the Black & Scholes model is meaningless since it models the underlying as being log normal. This is why practitioners have resorted to the displaced log normal model, where the variable $underlying + displacement$ is assumed to be log normal, with $displacement$ an additional parameter of the model.


in interest rates , the logN model is not correct since rates can be negative. so you should use the Normal (Gaussian) model, this is the standard now.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.