In the binomial model, how that the Delta of a call option $\Delta^{call}$ and the Delta of a put option $\Delta^{put}$ with the same maturity and strike satisfy $$\Delta^{call}_t - \Delta^{put}_t = 1, \ \ \text{for all} \ \ t = 0,\ldots, T-1$$ Is this result model-independent? Hint: put-call parity.

Excuse the grammatical mistakes of the question, English is not my professors first language. I don't understand what he means in asking if the model is independent? Any suggestions is greatly appreciated.

  • $\begingroup$ Put call parity is independent of any model, so if you could use PCP.... you would have a model independent result, $\endgroup$ – Alex C Feb 11 '16 at 2:23
  • $\begingroup$ On the other hand if you prove a result starting from the Black-Scholes formula, then you have a Balck Scholes model dependent result, which is less general. $\endgroup$ – Alex C Feb 11 '16 at 2:30
  • $\begingroup$ So I am asked to prove that the PCP has an independent result? $\endgroup$ – Wolfy Feb 11 '16 at 2:44
  • $\begingroup$ By "result" I meant the statement $\Delta^{call}-\Delta^{put}=1$. This statement needs to be proved starting from PCP, without binomial or BS assumptions. $\endgroup$ – Alex C Feb 11 '16 at 11:45
  • $\begingroup$ @AlexC could you provide an answer I am not getting anywhere $\endgroup$ – Wolfy Feb 11 '16 at 17:08

First, we have $P(t)+S(t)=C(t)+B(t,T)\cdot K$,

Then, $\frac{\partial P(t)}{\partial S(t)} + \frac{\partial S(t)}{\partial S(t)} = \Delta^{\text{put}}_{t}+1$ and $\frac{\partial C(t)}{\partial S(t)} + \frac{\partial [B(t,T)\cdot K]}{\partial S(t)} = \Delta^{\text{call}}_{t}+0$. Finaly, $\Delta^{\text{call}}_{t}-\Delta^{\text{put}}_{t}=1$.

This relationship is model-free, in sense, to derive this result we didn't use that we are in binomial model :) So the result is model-idependant.

  • $\begingroup$ I believe we need to show the proof using the binomial model, the proof you provide (correct as it is) is using the black-scholes model I believe $\endgroup$ – Wolfy Feb 11 '16 at 16:55
  • $\begingroup$ Can we do for the binomial model? $\endgroup$ – Wolfy Feb 11 '16 at 17:49

I use only the fact that

$S(t)\cdot B^{-1}(t)$ is a $\mathbb{Q}-$ martingale and we are considering the European options.


We have that : $(x-K)_{+}-(K-x)_{+}=x-K$ $\forall x,K \in \mathbb{R}$ ($\diamond$)

But we know also that the price of a European Call/Put is given by :

$C(t) = \mathbb{E}^{\mathbb{Q}}\left[\left(S_T-K\right)_{+}\cdot \frac{B(t)}{B(T)}\, \mid \, \mathcal{F}_{t}\right]$ , $P(t) = \mathbb{E}^{\mathbb{Q}}\left[\left(K-S_T\right)_{+}\cdot \frac{B(t)}{B(T)}\, \mid \, \mathcal{F}_{t}\right]$

Using the ($\diamond$), we have $C(t)-P(t) = \mathbb{E}^{\mathbb{Q}}\left[\left(S_T-K\right)\cdot \frac{B(t)}{B(T)}\, \mid \, \mathcal{F}_{t}\right]=B(t)\cdot \mathbb{E}^{\mathbb{Q}}\left[ \frac{S_T}{B(T)}\, \mid\,\mathcal{F}_{t}\right]-\frac{B(t)}{B(T)}\cdot K$

Using the assumption, we have $B(t)\cdot \mathbb{E}^{\mathbb{Q}}\left[ \frac{S_T}{B(T)}\, \mid\,\mathcal{F}_{t}\right]=B(t)\times \frac{S_{T}}{B(t)} = S_{t}$

We also know that $\frac{B(t)}{B(T)}=B(t,T)$ (I think if not, there's arbitrage)

At the end, we get the result :)


In the binomial model suppose the stock can go to either U or D. Suppose the option strike is K where D < K < U.
If stock goes to U, call payout is (U-K) and put payout is 0. If stock goes to D, call payout is 0 and put payout is (K-D). Now consider a portfolio of Call - Put. If stock goes to U, portfolio payout is U-K. If stock goes to D, portfolio payout is D-K. Therefore, portfolio is the same as stock - K (ie a forward on the stock). Therefore its delta is one, in the binomial model. Of course its delta is one in any model, as others have pointed out.


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.