Chapter 27 · Exotic Options

Barrier Options & Path Dependency

When the journey matters as much as the destination — the Reflection Principle, joint densities, and the trapdoor option.
The journey now matters

All of Chapters 13–25 priced options that only cared about where the stock was at expiry. European calls, puts, and even Asian options (once we applied the running-sum trick) depend on the terminal state or a function of the full path. Exotic options introduce a harder requirement: the price can be killed, amplified, or restructured by what happens along the way. The maximum, the minimum, or a single barrier crossing can change everything.

Vanilla (Chapters 13–25)

Payoff depends only on S(T) — the final price. The path taken is irrelevant. Markov property makes pricing tractable. BSM PDE with standard boundary conditions.

Payoff = h(S(T)) only

Exotic (Chapter 27)

Payoff depends on M(T) = max S(u), or whether S ever touched a barrier B. The entire path history is embedded in the price. Requires joint densities and stopping times.

Payoff = h(S(T), M(T)) — path matters
Vanilla vs barrier option — same final price, different outcome because one path hit the barrier Two stock price paths shown. Both start at S0 and end at the same final price above the strike K. But the second path crosses a barrier B during the life of the option. The vanilla option pays out for both paths. The barrier option only pays out for path 1 — path 2 was knocked out at the barrier. B (barrier) K (strike) Path 1 ✓ Pays out KNOCKED OUT here! Path 2 (ghosted) ✗ Option = 0 Same S(T)! t = 0 T →

Both paths end at the same price above the strike. The vanilla option pays out for both. The Up-and-Out barrier option pays only for Path 1 — Path 2 was killed the instant it touched B.

1 · The reflection principle — the mathematical mirror

How do we compute the probability that a Brownian path never crosses a level B? The direct calculation is formidable — you need to exclude an uncountable family of paths. The Reflection Principle provides an elegant shortcut by exploiting a symmetry of Brownian motion: once a path hits level B, it is equally likely to continue above or below. This lets us "count" the paths that crossed B by reflecting them.

The Reflection Principle — Step by Step
1
A Brownian path W(t) hits barrier B at some time τ (the stopping time — the first crossing).
2
After time τ, reflect the path through B: the reflected path W̃(t) = 2B − W(t) for t ≥ τ. It is a valid Brownian path by symmetry.
3
Key insight: every path that ends above B corresponds bijectively to a reflected path that ended below B but crossed B at some point. The probability masses are equal.
4
Therefore: P(ever hits B AND ends at w) = P(ends at 2B − w) — the mirror image. This converts a complex path-integral into a simple terminal distribution calculation.
Reflection principle — a path hitting barrier B is reflected through B to create a mirror path A Brownian path rises to hit barrier B at time tau, then the portion after tau is reflected through B to create the mirror path. The original path ends at w above B, and the reflected path ends at 2B minus w below B. B τ (first hit) w (original endpoint) 2B−w (reflected) mirror t = 0 T → P(hits B AND ends at w) = P(ends at 2B−w)

The purple path hits B at time τ and ends at w above B. The green reflected path ends at 2B−w below B. These events have identical probability — the mirror symmetry of Brownian motion.

The Reflection Principle converts a question about paths that never cross a level into a question about terminal values at the mirror level. Terminal distributions of Brownian motion are just Gaussian — easy to compute. The reflection trick is why barrier option pricing has closed-form solutions at all.

2 · The joint density — tracking price and maximum simultaneously

To price path-dependent options, we need to track two things at once: where the stock ends up (S(T) = w) and the highest level it ever reached (M(T) = m). The Reflection Principle produces the joint density of these two quantities — the probability of both happening simultaneously.

Theorem 27.1 — Joint Density of (W(T), M(T))
f_{W,M}(w, m) = (2(2m−w) / T√(2πT)) · exp{−(2m−w)² / 2T}
Valid for: w ≤ m and m ≥ 0 (the maximum must be at least as large as the terminal value)

Two conditions encoded simultaneously:
The path ended at w  AND  the highest point ever reached was exactly m

Derived directly from the Reflection Principle: P(W(T)∈dw, M(T)∈dm) = this formula × dw dm

The joint density is the mathematical engine behind both barrier and lookback options. For a barrier option: integrate over all w < B and m < B (paths that never touched the barrier). For a lookback option: integrate over all w with the maximum m treated as the "asset" being paid off. The same formula, different limits of integration.

3 · The up-and-out call — the trapdoor option

The most common barrier exotic is the Up-and-Out Call. It behaves exactly like a regular call option — until the stock touches an upper barrier B, at which point the option instantly expires worthless. The buyer pays less premium than for a vanilla call because they accept the knockout risk. The seller collects the premium but must carefully hedge the barrier.

ConditionMathLayman explanation
Lower bound v(t, 0) = 0 If the stock hits zero, the call is dead — a bankrupt company pays nothing. Same as vanilla BSM.
Upper bound ★ v(t, B) = 0 The critical new condition. The moment the stock touches the barrier B, the option value drops to exactly zero — instantly. This is the "trapdoor."
Terminal v(T, x) = (x−K)⁺ · 𝟙{x < B} At expiry, if you survived (never hit B), collect the vanilla payoff. If you previously hit B, you already received zero. The indicator 𝟙{x<B} enforces this.
PDE (interior) vₜ + rxvₓ + ½σ²x²vₓₓ = rv Same Black-Scholes PDE as always — only the boundary conditions change. The interior dynamics are identical to vanilla options.
Up-and-out call option value surface — value collapses to zero at the barrier boundary A 3D-style surface showing call option value. The surface rises with stock price below the barrier but drops abruptly to zero at the barrier B. Near expiry the surface becomes discontinuous at B. B (barrier) K (strike) v = 0 beyond B Near expiry (sharper cliff) v(t,x) x = 0 x →

The up-and-out call value surface: rises with stock price but drops to zero at the barrier B. Near expiry (gold dashed) the cliff becomes sharper — causing the Greeks to explode.

4 · The up-and-out pricing formula — four parts

Solving the BSM PDE with the barrier boundary conditions produces a formula with four terms. The first two look like a standard BSM call. The last two are "correction" terms that subtract the value of paths that would have been profitable but were knocked out by crossing the barrier.

Up-and-Out Call Formula (Theorem 27.2)
v(t,x) = [c_BSM(x)] − [correction for paths that hit B]

= xN(d₁(x)) − Ke⁻ʳᵀN(d₂(x))
− x(B/x)²λN(d₁(B²/x)) + Ke⁻ʳᵀ(B/x)^{2λ−2}N(d₂(B²/x))
Term 1 — xN(d₁)
Standard BSM call, first part. Expected stock price weighted by probability of exercise. Same as vanilla.
Term 2 — Ke⁻ʳᵀN(d₂)
Standard BSM call, second part. Discounted strike weighted by exercise probability. Same as vanilla.
Term 3 − x(B/x)²λN(d₁(B²/x))
First correction. Subtracts the "reflected" paths — those that hit B and would have continued profitably. λ = (r + ½σ²)/σ².
Term 4 + Ke⁻ʳᵀ(B/x)^{2λ−2}N(d₂(B²/x))
Second correction. Adds back the discount side of the reflected paths to keep the formula balanced. Uses B²/x as the "mirror price."

The key insight: B²/x is the "mirror price" — the reflection of stock price x through the barrier B on the log scale. Every term in the vanilla BSM formula has a corresponding reflected term using B²/x instead of x. This is the Reflection Principle made algebraic — and it produces a closed-form answer for a genuinely path-dependent derivative.

5 · The delta-hedging nightmare near the barrier

Barrier options create a practical hedging problem that has no equivalent in vanilla options: as the stock approaches the barrier B near expiry, the option value must drop from a positive number to exactly zero in a vanishingly small price interval. This forces the Greeks to explode.

⚡ The Greeks Explosion Near the Barrier
Delta (Δ = vₓ): As x → B near T, Delta must go from its vanilla value to −∞ to force v to zero at the boundary. A huge negative delta means: sell massive amounts of stock.

Gamma (Γ = vₓₓ): Gamma explodes in the opposite direction — the curvature becomes enormous. Small stock movements near the barrier require massive hedge rebalancing.

The Practical Fix: Quants "shift the barrier" — they price and hedge as if the barrier were slightly higher than B (e.g., B + 2σ√dt). This prevents infinite trading while staying close to the correct hedge. In practice, barrier options on liquid underlyings are hedged with a portfolio of vanilla options at multiple strikes near B, not just stock.
Delta explosion near barrier — delta becomes very large negative as stock approaches barrier near expiry A chart showing the up-and-out call delta as a function of stock price. Far from the barrier, delta looks like a vanilla call delta between 0 and 1. Near the barrier, delta drops sharply to a large negative value then returns to zero at the barrier. 0 B Vanilla Δ Barrier Δ Δ → −∞ ! S = 0 S → B Delta explodes negative near barrier — massive short position required

Barrier delta (gold) looks like vanilla delta (blue dashed) far from B — but explodes to −∞ just before the barrier near expiry. In practice, this is tamed by shifting the hedge barrier or using a vanilla option replication portfolio.

Try it — barrier option simulator

Simulate many GBM stock paths. Count which ones survive (never touch barrier B) and compute the up-and-out call price via Monte Carlo. Compare to the analytical formula. Watch what happens to the survival rate as the barrier moves closer to the current stock price.

$100
$100
$130
25%
Vanilla call
Barrier call (MC)
Survival rate
Knockout discount

Green paths = survived (never hit barrier)  |  Red paths = knocked out at barrier B  |  Gold dashed = barrier level  |  Barrier call is always cheaper than the vanilla call

Chapter 27 — exotic options and barrier pricing in five ideas:

What comes next: The Lookback Option — where the payoff is max(S(T) - min{S(u)}, 0), paying you the "best possible" profit you could have made with perfect timing. The same joint density, integrated with different limits.