Constant-Depth Quantum Barrier Option Pricing via a Static Bivariate Oracle Framework

AI summary

Overview

Research area: Quantum computing for computational finance — specifically gate-based quantum algorithms for pricing path-dependent derivative contracts (barrier options) using quantum amplitude estimation.

Technical level: Advanced. The paper assumes familiarity with stochastic calculus (Brownian motion, Girsanov's theorem, the reflection principle), quantum circuit design (ripple-carry comparators, multi-controlled rotations, uncomputation), and quantum amplitude estimation variants (QAE, MLAE, IQAE).

Scope: The paper proposes and benchmarks a constant-depth quantum oracle that prices barrier options by encoding the joint distribution of the terminal price and the running supremum, replacing sequential monitoring-date circuits with a fixed two-register architecture.

What This Paper Is About

Existing quantum algorithms for barrier options track the asset price at every monitoring date, so the circuit grows linearly in qubit count and two-qubit gate depth as the number of dates d increases. That makes high-frequency contracts such as daily monitoring (d = 252) impractical on real hardware, and prior experimental work has been confined to stylized settings such as d = 2. The authors' goal is to remove the temporal dependency entirely, so that circuit cost is set by the desired pricing resolution rather than by how often the barrier is observed.

Key Contributions

1. Closed-form bivariate distribution via Brownian reflection. Combining the Brownian reflection principle, the Girsanov measure change, and a Jacobian transform back to spot coordinates, the authors derive an exact joint probability density f_{S_T,M_T}(s,m) for the terminal price and the running supremum, supported on the wedge domain where m ≥ S_0 and 0 < s ≤ m. This replaces temporal discretization with an exact classical evaluation.

2. Constant-depth static bivariate oracle. The dynamic boundary-crossing check is replaced by two parallel single-condition comparators (checking M_T ≥ B_adj and S_T ≥ K) plus one dual-controlled rotation gate. The authors prove analytically that both circuit depth and comparator count drop from O(d) to O(1) with respect to monitoring frequency.

3. Zero-quantum-cost discrete monitoring correction. The Broadie–Glasserman–Kou continuity correction is absorbed into classical preprocessing by shifting the barrier to B_adj = B · exp(±βσ√Δt) with β ≈ 0.5826, so daily monitoring is handled without any additional quantum gates.

4. Transpilation and noise benchmarks. Under Qiskit transpilation to a superconducting gate set {u1, u2, u3, cx}, the framework reduces physical CNOT count by 98.59% (2262 to 32) and two-qubit depth by 98.02% (1514 to 30 layers) at d = 252, and retains a 7.84% relative valuation error under a 127-qubit IBM Eagle noise model with M3 readout mitigation.

Main Findings

• Comparator and depth reduction: The baseline sequential architecture requires d + 1 full-precision comparators and a cascading OR-tree of d − 1 multi-qubit Toffoli gates with d − 2 borrowing ancillae; the proposed circuit requires exactly two static comparators and a single dual-controlled linear rotation.

• Qubit footprint decoupling: The baseline allocates d · n_step + 2d + n_step qubits, i.e. O(d · n_step). The proposed footprint is n_s + n_m + 3 + max(n_s, n_m), which is O(n_s + n_m), stated as O(1) with respect to d.

• Physical gate counts at d = 252: CNOT count falls from 2262 to 32 (98.59% reduction) and two-qubit circuit depth from 1514 to 30 layers (98.02% reduction) after transpilation to the superconducting gate set.

• Noise resilience: Under a realistic 127-qubit IBM Eagle device noise model, the shallow circuit achieves a relative valuation error of 7.84% when combined with M3 readout error mitigation.

• State preparation overhead is bounded: With the benchmark configuration n_s = n_m = 4 (8 coordinate qubits total), Möttönen-style recursive uniformly controlled rotation decomposition needs at most 255 CNOTs, a budget the authors note can be compressed via classical Schmidt decomposition or shallow qGANs. Because n_s and n_m are fixed by pricing tolerance rather than monitoring schedule, this cost is O(1) in d.

• In-out parity is preserved exactly: Summing the up-and-in and up-and-out measurement probabilities reproduces the vanilla price at the quantum state level; the up-and-out case requires only a single Pauli-X on the barrier flag, not a redesigned oracle.

• Discrete-to-continuous bridging: Shifting the barrier threshold by the BGK factor β ≈ 0.5826 (derived from the Riemann zeta function as β = −ζ(1/2)/√(2π)) neutralizes discrete sampling bias entirely in classical preprocessing.

• Extension beyond closed-form contracts: The authors report that the framework extends to nonlinear exotic options lacking closed-form analytical solutions, though the truncated content provided does not give the specific contracts, parameters, or error figures for that demonstration.

• Not reported in the available content: The explicit closed-form optimal scaling parameter c*(M), the numerical breakdown of the grid / Taylor truncation / statistical error budget, and the full Section 5 numerical tables are referenced but their values do not appear in the text supplied.

Methodology in Plain English

The central move is to stop simulating the asset's journey and instead describe the endpoint of that journey statistically. For a barrier option, the payoff depends on only two numbers: where the price ends up at expiry, and the highest level it ever reached along the way. Classical probability theory gives an exact formula for how those two numbers are jointly distributed, obtained by reflecting price paths that cross a level and then correcting for the drift of the price process. The authors import that formula into a quantum circuit.

Concretely, they build two quantum registers, one holding a discretized terminal price and one holding a discretized running maximum, and prepare them in an entangled superposition weighted by the exact joint density. The domain is truncated using a 3σ confidence envelope. Two arithmetic comparators then ask two static yes/no questions — did the maximum exceed the adjusted barrier, and is the terminal price above the strike — and a single rotation on a payoff qubit fires only when both answers are yes. The rotation angle is set so that the probability of measuring the payoff qubit in state 1 is linearly related to the expected payoff, after which amplitude estimation recovers the value and classical affine inversion converts it to a discounted price.

Two details make the construction practical. First, the number of qubits in each register depends on how finely the price range is gridded, not on how many days the barrier is observed, so the circuit does not grow with monitoring frequency. Second, the discrepancy between continuous and daily monitoring is fixed by moving the barrier slightly, an adjustment performed on the classical integer fed into the comparator, costing nothing in quantum gates.

Why This Matters

Impact on research. The paper reframes path dependence as a static distributional problem rather than a time-stepping simulation. If the argument holds, the standard objection to quantum barrier pricing on near-term hardware — that coherence cannot survive thousands of layers of two-qubit gates — is substantially weakened, and the practical ceiling on monitoring frequency is set by classical discretization tolerance instead. It also gives the quantum finance community a concrete, transpiled resource comparison at an institutionally realistic monitoring frequency, rather than an asymptotic claim.

Real-world applications:

• Pricing and hedging exchange-traded and OTC knock-in / knock-out barrier structures, which are commonly monitored at daily closes over annual horizons.
• Overnight portfolio stress testing and Value-at-Risk / Expected Shortfall runs across large books of path-dependent contracts, where Monte Carlo convergence at O(M^(−1/2)) is the bottleneck.
• Real-time intraday risk revaluation, where the shallow circuit's coherence resilience under device noise is directly relevant.
• Extending to nonlinear exotic payoffs without closed-form solutions, where the classical alternative is brute-force simulation.

Industry relevance. Institutional barrier agreements overwhelmingly specify high-frequency discrete monitoring, so a method whose cost is independent of monitoring frequency addresses the actual contract terms in use rather than a stylized approximation. The explicit CNOT and depth figures under a named hardware gate set, plus a named device noise model (127-qubit IBM Eagle) and a named mitigation technique (M3), make the claims auditable by practitioners evaluating near-term feasibility.

Future Directions

1. Quantify the nonlinear-exotic extension. The paper states the framework extends to nonlinear exotic options lacking closed-form solutions but the truncated content gives no contract specifications or error figures; reproducing and benchmarking those cases is the natural next step.

2. Close the error budget empirically. The authors propose decomposing total pricing bias into grid, Taylor truncation, and statistical estimation components and deriving a closed-form optimal scaling parameter c*(M); validating those analytical predictions against simulated device data is open.

3. Reduce state-preparation cost on hardware. The Möttönen decomposition's bound of 255 CNOTs at n_s = n_m = 4 is the dominant remaining circuit cost; the authors point to classical Schmidt decomposition and shallow qGANs as compression routes, which need demonstration.

4. Test on real silicon beyond noise emulation. All reported robustness results come from emulation under a 127-qubit IBM Eagle noise model rather than execution on a physical device, and multi-barrier or multi-asset generalizations are not addressed.

Target Audience

Quantum algorithm researchers working on financial applications and amplitude estimation; quantum finance practitioners evaluating hardware feasibility for path-dependent products; and quantitatively trained derivatives professionals or computational finance researchers who want to understand what quantum speedups currently look like at realistic monitoring frequencies. Readers need comfort with stochastic calculus and gate-level circuit reasoning to follow the derivations, though the resource comparisons and benchmark figures are accessible to a broader quantitative audience.

Authors’ abstract

Quantum amplitude estimation offers a theoretical quadratic speedup over classical Monte Carlo methods for financial derivative pricing. However, extending gate-based algorithms to path-dependent barrier options has historically been hampered by sequential path discretization, where tracking price paths across $d$ monitoring dates incurs a linear escalation in qubit footprint and gate depth ($\mathcal{O}(d)$), posing severe coherence challenges for near-term quantum processors. In this work, we present a constant-depth quantum pricing framework that fundamentally eliminates the temporal dependency of barrier monitoring. Leveraging the Brownian reflection principle and Girsanov theorem, we project the continuous supremum dynamics onto a static bivariate joint distribution of the terminal price $S_T$ and running supremum $M_T$. This allows decoupling dynamic boundary checks into a static bivariate oracle comprising two parallel comparators and a dual-controlled rotation gate. We prove this architecture reduces circuit depth and comparator complexity from $\mathcal{O}(d)$ to an invariant $\mathcal{O}(1)$, while supporting discrete monitoring via zero-quantum-cost boundary corrections. Transpilation benchmarks for daily monitoring ($d = 252$) show our framework slashes physical CNOT count by 98.59% (from 2262 to 32) and compresses two-qubit depth by 98.02% (from 1514 to 30 layers). Under a realistic 127-qubit IBM Eagle noise model, the shallow circuit achieves a 7.84% relative valuation error with M3 readout mitigation. Finally, the framework seamlessly extends to nonlinear exotic payoffs lacking analytical solutions, establishing a noise-resilient pathway toward practical quantum advantage in computational finance.