1
$\begingroup$

Consider a derivative of digital type which pays this kind of payoff at time $T$: \begin{align*} g(S_T,k) &= \begin{cases} P_0,~S_T>k \\ S_T, ~S_T\leq k \end{cases} \end{align*}

with $S_T$ being the current price of the underlying at maturity time $T$, $P_0$ the price of the underlying at the issue time 0 and $k$ - kind of the strike price with barrier feature.

Apparently, function $g$ is discontinuous at $S_T=k$ and has a jump there. The idea is to approximate it with a set options, call $c(S_T,k_1)$ and put $p(S_T,k_2)$ that have strikes: $k_1 < k < k_2$. Then, to construct a linear piece-wise function that will look as following: $$ \hat g(S_T,k_1,k_2)=a_0+a_1 S_T+a_2 c(S_T,k_1) + a_3 p(S_T,k_2). $$

The question is how to get the coefficients. Which complementary equations may be used?

$\endgroup$
  • 3
    $\begingroup$ Note that \begin{align*} g &= P_0 \pmb{1}_{S_T > K} + S_T \pmb{1}_{S_T\le K}\\ &= P_0 \pmb{1}_{S_T > K} + S_T \left(1-\pmb{1}_{S_T> K}\right)\\ &=(P_0-K)\pmb{1}_{S_T > K} + S_T - (S_T-K)^+. \end{align*} $\endgroup$ – Gordon May 17 at 15:44
  • $\begingroup$ Moreover, $$\pmb{1}_{S>K} \approx \frac{(S_T-(K-\varepsilon))^+-(S_T-K)^+}{\varepsilon},$$ for a sufficiently small $\varepsilon$ (e.g., $10^{-4}K)$. $\endgroup$ – Gordon May 18 at 11:36
2
$\begingroup$

We should be able to replicate the payoff exactly in each of the two regions $S_{T}\leq k_{1}$ and $S_{T}\geq k_{2}$. From the first, $$a_{0}+a_{1}S_{T}+a_{3}(k_{2}-S_{T}) =S_{T}$$ so, matching coefficients, $a_{0}+a_{3}k_{2}=0$ and $a_{1}-a_{3}=1$. From the second, $$a_{0}+a_{1}S_{T}+a_{2}(S_{T}-k_{1})=P_{0}$$ so, matching coefficients, $a_{0}-a_{2}k_{1}=P_{0}$ and $a_{1}+a_{2}=0$.

Sorry I haven't time to check this works. Hope it helps.

$\endgroup$
  • $\begingroup$ Solved by $a_{0}=(P_{0}-k_{1}) k_{2}/(k_{2}-k_{1})$, $a_{1}=(k_{2}-P_{0})/(k_{2}-k_{1})$, $a_{2}=(P_{0}-k_{2})/(k_{2}-k_{1})$ and $a_{3}=(k_{1}-P_{0})/(k_{2}-k_{1})$. It would probably come out a bit nicer if you swapped the put for a call at the same strike. $\endgroup$ – Ali May 18 at 12:36
  • $\begingroup$ Thank you! This does work! $\endgroup$ – harve May 19 at 0:02

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.