Breenden-Litzenberg

Breenden-Litzenberg#

General#

Consider call price \(C(K) = (S-K)^+\), where \(S\) is the stock price at expiry

We clearly have

\[ E_Q[ (S - K)^+] = (S-K) \cdot P(S > K) \]

And

\[\begin{align*} \frac{\partial}{\partial K} C(K) =& \frac{\partial}{\partial K} \mathbb{E} \left( (S(T) - K)^+ \right) \\ =& \frac{\partial}{\partial K} (S-K) \cdot P(S > K)\\ =& - P(S > K) \end{align*}\]

In other words, the probability that the call option expires ITM, is simply

\[\begin{align*} P(S > K) = - \frac{\partial C}{\partial K} \end{align*}\]

Denote the CDF of S as \(F_S(x) = P(S \leq x)\), then

\[\begin{align*} P(S \leq K) = F_S(K) = 1+ \frac{\partial C}{\partial K} \end{align*}\]

Differentiating again we have

\[\begin{align*} f_S(K) =& \frac{\partial^2}{\partial K^2} C(K) \end{align*}\]

Which is the market implied density of the stock. Similarly for puts:

\[\begin{align*} \frac{\partial Put(K) }{\partial K} =& \frac{\partial}{\partial K} \mathbb{E} \left( (K - S(T))^+ \right) \\ =& \frac{\partial}{\partial K} (K - S) \cdot P(S \leq K)\\ =& P(S \leq K) \end{align*}\]

And

\[\begin{align*} \frac{\partial^2}{\partial K^2} p(K) = f_S(K) \end{align*}\]

So in conclusion:

\[\begin{align*} P(S \leq K) =& \frac{\partial P}{\partial K} = 1+ \frac{\partial C}{\partial K} \\ f_S(K) =& \frac{\partial^2}{\partial K^2} C(K) = \frac{\partial^2}{\partial K^2} P(K) \end{align*}\]

Black Scholes Framework#

Under BS, we have

\[ C = S_0 \Phi(d_1) - K e^{-rT} \Phi(d_2) \]
\[ d_1 = \frac{\ln\left(\frac{S_0}{K}\right) + \left(r + \frac{\sigma^2}{2}\right)T}{\sigma \sqrt{T}}, \quad d_2 = d_1 - \sigma \sqrt{T} \]

Considering only near-dated options, we can assume \(r = 0\), we have

\[ \frac{\partial C}{\partial K} = -\Phi(d_2) \]

so that

\[\begin{align*} P(S > K) = - \frac{\partial C}{\partial K} = \Phi(d_2) \end{align*}\]

Note that delta is given by \(\Phi(d_1)\) which is very close to \(\Phi(d_2)\), as \(T\) is very small for near-dated options.

Hence for short dated ATM call options tend to have delta = 0.5, as that is its N(d1) = N(d2) = probability expiring ITM.

i.e. For ATM (\(S = K\)), as \( T \to 0\):

\[ d_{1,2} = \frac{\left(r \pm \frac{\sigma^2}{2}\right)T}{\sigma \sqrt{T}} \to \frac{\sigma \sqrt{T}}{2} \to 0 \]
\[ \phi(d_1), \phi(d_2) \to 0.5 \]