11
$\begingroup$

I’m studying the BSM model and having a look at the greeks. I was reading Derivatives, by Paul Wilmott, and he gives the closed form solutions without making the reader see where these solutions come from.

Is there a good book that explains in detail how each greek, from Delta to Rho, is derived and computed?

$\endgroup$
  • 3
    $\begingroup$ The real question about the Greeks these days is not where do they come from but where do they go to... (sorry, couldn't resist :-) $\endgroup$ – vonjd Jun 29 '15 at 16:20
  • 1
    $\begingroup$ I think some interview prep books like "A Practical Guide to Quantitative Finance Interviews" by Zhou have such a derivation. $\endgroup$ – bcf Jun 29 '15 at 23:28
  • 1
    $\begingroup$ I've gave a read at the book you've suggestested. Even there the greeks are "given", instead I want to know if I can compute them from a PDE with known boundary conditions. $\endgroup$ – james42 Jun 30 '15 at 0:57
6
$\begingroup$

First, my notation. $K$ is the strike price, $S$ is the stock price, $r$ is the continuously compounded risk-free rate, $T$ is time at expiration, $t$ is time at issue, $\sigma$ is volatility, $\delta$ is continuously compounded dividend rate.

The Black-Scholes formula for a European call is

$C = Se^{-\delta (T-t)} N(d_1) - Ke^{-r(T-t)} N(d_2)$

$d_1 = \dfrac{\ln(S/K) + (r - \delta + 0.5\sigma^2)(T-t)}{\sigma \sqrt{(T-t)}}$ and $d_2 = d_1 - \sigma \sqrt{(T-t)}$.

Some common greeks are

$\Delta$ = $\dfrac{ \partial C}{\partial S}$, $\Gamma = \dfrac{ \partial^2 C}{\partial S^2}$, $\rho = \dfrac{\partial C}{\partial r}$, $v = \dfrac{\partial C}{\partial \sigma}$, $\theta = \dfrac{\partial C}{\partial t}$ and $\psi = \dfrac{\partial C}{\partial \delta}$.

Note that $\theta$ is often equivalenty defined as $- \dfrac{\partial C}{\partial T}$. You can derive the greeks by taking the partial derivatives.

As an example, I will derive $\Delta = e^{-\delta (T-t)} N(d_1)$

\begin{align*} \Delta &= \dfrac{\partial C}{\partial S} \\ &= e^{-\delta (T-t)} N(d_1) + \dfrac{\partial C}{\partial S} Se^{-\delta (T-t)} N(d_1) - \dfrac{\partial C}{\partial S} Ke^{-r(T-t)} N(d_2) \\ &= e^{-\delta (T-t)} N(d_1) \end{align*}

It is not obvious that the last two terms cancel out. I prove this below.

\begin{align} &\dfrac{\partial C}{\partial S} Se^{-\delta (T-t)} N(d_1) - Ke^{-r(T-t)} N(d_2) \\ &= Se^{-\delta (T-t)} \dfrac{\partial}{\partial S} d_1 \dfrac{1}{\sqrt{2\pi}} e^{-0.5d_1^2} - Ke^{-r(T-t)} \dfrac{\partial}{\partial S} d_2 \dfrac{1}{\sqrt{2\pi}}e^{-0.5d_2^2} \\ &\propto \dfrac{\partial}{\partial S}d_1 \left( Se^{-\delta (T-t)} e^{-0.5d_1^2} - Ke^{-r(T-t)} e^{-0.5 d_2^2} \right) \\ \end{align}

Note that

\begin{align*} &\ln\left(Se^{-\delta(T-t)}e^{-0.5d_1^2}\right) - \ln\left(Ke^{-r(T-t)} e^{0.5d_2^2} \right) \\ &= \ln(S) - \delta(T-t) - \ln(K) + rT + 0.5d_2^2 - 0.5d_1^2 \\ &= \ln(S/K) + (r-\delta)(T-t) - 0.5(d_1^2 - (d_1 - \sigma\sqrt{T-t})^2) \\ &= \ln(S/K) + (r-\delta)(T-t) - 0.5(2d_1\sigma\sqrt{T-t} - \sigma^2(T-t)) \\ &= \ln(S/K) + (r-\delta)(T-t) - d_1 \sigma\sqrt{T-t} + 0.5\sigma^2(T-t) \\ &= \ln(S/K) + (r-\delta)(T-t) - \left(\ln(S/K) + (r-\delta + 0.5\sigma^2)(T-t)\right) + 0.5\sigma^2(T-t) \end{align*}

So $\left( Se^{-\delta (T-t)} e^{-0.5d_1^2} - Ke^{-r(T-t)} e^{-0.5 d_2^2} \right) = 0$ and $\Delta$ = $e^{-\delta(T-t)} N(d_1)$ as shown above.

Note that $\Delta > 0$ for a European call. Hedging greeks is a common topic in financial economics. To hedge a European call, short sell $\Delta$ shares of stock. This protects a portfolio against small changes in the stock price.

EDIT 1 By the BSE, $V_t + 0.5V_ss(\sigma S)^2 = rV - VsS(r-\delta)$ can be written as $\theta + 0.5\Gamma(\sigma S)^2 = rV - \Delta S(r-\delta)$. The solution to the BSE depends on the terminal conditions and payoff.

$\endgroup$
  • 1
    $\begingroup$ Easy as a pie. I want to know deeper if we can see the greeks as closed form solutions of PDE written by imposing some conditions... $\endgroup$ – james42 Jun 29 '15 at 15:50
3
$\begingroup$

The following paper gives a simple derivation of the BSM (via a simple integration approach instead of the classical PDE approach) and the Greeks plus some intuition for each:

Derivation and Comparative Statics of the Black-Scholes Call and Put Option Pricing Formulas by Garven, J.

You find the derivation of the Greeks in chapter 4 (called "comparative statics") on p. 12ff.

$\endgroup$
  • $\begingroup$ Thank you. But I've takled the problem in a different way: if we take the original BS PDE, then differentiate it w.r.t. to $r$ for example, we can prove that the solution to this PDE is exactly the greek rho. The problem is that I don't know how the boundary conditions in this case would change. $\endgroup$ – james42 Jul 5 '15 at 15:20
  • $\begingroup$ @ale42: I answered your question about references how to derive the Greeks. Why do you ask something when you mean something else? $\endgroup$ – vonjd Jul 5 '15 at 15:25
  • $\begingroup$ Maybe in the main post wasn't so clear, but in my comment below I wrote that I'd preferred using a PDE-based approach because I was looking for mathematical insights. Maybe next time I'll explicit better my needs. ;) $\endgroup$ – james42 Jul 5 '15 at 15:37

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.