Approximating Digital Call Payoffs with Call Spreads and Hedges
Summary
The discussion explains why a cash-or-nothing digital call cannot generally be replicated exactly using ordinary calls and puts at strikes spaced five dollars apart. In theory, a digital payoff can be approached by a call spread whose strike gap shrinks toward zero; with a finite gap, the spread is only an approximation. A static spread may work better when the chance of the underlying finishing within the gap is small, but its payoff and sensitivity diverge near the digital strike.
One proposed improvement is to dynamically delta hedge the spread to manage the mismatch near the strike, though liquidity can make frequent adjustment impractical. The answers also construct upper and lower price bounds using scaled call spreads on either side of the target strike. These bounds reflect the payoff shortfall or excess in the interval between strikes. The discussion is conceptual: actual replication quality and bounds depend on available strikes, market liquidity, and the underlying's distribution.
Key ideas
- A digital call can be approximated by a call spread as the distance between its strikes narrows.
- A spread with a finite strike gap does not reproduce the discontinuous digital payoff exactly.
- Dynamic delta hedging can address some mismatch near the strike, subject to liquidity constraints.
- Scaled spreads above and below the target payoff can provide theoretical price bounds.
Tags
Full text
# How to replicate a digital call option
# How to replicate a digital call option
Call Option S=100 K=100 Payoff=1 (option is not available) How can i replicate this (payoff) with calls and puts with strike prices with multiples of 5$
Thanks for help
## Answer by SmallChess (score 13)
https://quant.stackexchange.com/a/18938
A digital call option (cash-or-nothing) can be replicated with two call options with different Strike. When we make the delta infinitely small and assume we have arbitrary strike prices. We get:
$C_{\textsf{bin}}(K,t) = \lim_{h \to 0} \frac{C(K,t)-C(K+h,t)}{h} = -\frac{\partial C(K,t)}{\partial K}$
## Answer by glyphard (score 3)
https://quant.stackexchange.com/a/1465
use a vertical spread and delta hedge it.
http://www.wilmott.com/messageview.cfm?catid=3&threadid=65988
## Answer by Mats Lind (score 0)
https://quant.stackexchange.com/a/49401
Going back to the original question, there is no static replication. This is clear from the first answer above which states that a call spread with an infitesimal difference between their strikes is needed. To arrive at an approximate replication, we need the probability of the underlying fixing in the USD 5 interval between the calls to be small, hence:
- Far away from the binary's strike, at large volatilities and long expiries, the call-spread replication could be good enough.
Going towards the strike at limited volatlity and time to expiry, the delta of the binary rises above that of the replica. Adjusting the positions in the options of the replica would probably not liquidity-wise be feasible - hence the second answer:
- Add a dynamic delta hedge to cover the delta mismatch near the strike of the binary.
## Answer by siou0107 (score 0)
https://quant.stackexchange.com/a/80100
Short answer: you cannot replicate exactly the digital option because its payoff function is discontinuous, making delta-hedging impossible close to the strike. However, you can over-hedge it: buying a static portfolio, payoff of which will dominate the digital payoff function at expiry.
Buy 0.2 call spread with strikes 95\$ and 100\$. When your digital option ends up in the money, the call spread payout will be 1; if it ends up out, you may get some extra money. Because that portfolio pays off more or the same than your digital, its value is theoretically higher than the digital's. It therefore gives you an upper bound on the price you should charge for the digital.
An equivalent lower bound is given by that of 0.2 100\$-105\$ call spread; that portfolio is worth the same as the digital on $\left[105, + \infty\right)$ and on $\left(-\infty, 100\right]$, but less than it on $\left(100, 105\right)$.
In your market conditions, you cannot do better. Any reliable model should give you a price for the digital option within those bounds.Shown in full with attribution under the source's licence. Licence: CC BY-SA 4.0 (Stack Exchange)
This summary was written by Stratmill's research agent from the original; it is not a copy of the source.