Toward Quantum Enabled Solutions for Real-Time Currency Arbitrage in Financial Markets

Suman Kumar Roy, Rahul Rana, M Girish Chandra, Niraj Kumar, Manoj Nambiar2025 · DOI

TL;DR

This paper tackles the challenge of real-time currency arbitrage by proposing an enhanced mathematical model that guarantees valid trading cycles. They benchmark quantum approaches like Quantum Annealing (QA) and Adaptive Cost Encoding (ACE) against classical solvers, finding that quantum methods offer competitive profit generation with significantly faster execution times, paving the way for their use in time-sensitive financial applications.

What Problem Does This Paper Address?

Currency arbitrage, the practice of exploiting tiny, fleeting price differences between currencies to make risk-free profits, is a computationally intensive problem. The vast number of possible trading routes (cycles) and the need for near-instantaneous decision-making make it a prime candidate for advanced computational techniques. Traditional methods often struggle with the speed and scale required, and prior attempts to model arbitrage have sometimes allowed for invalid or redundant trading cycles, leading to suboptimal or infeasible solutions. This paper aims to bridge this gap by developing a more robust model and evaluating the performance of emerging quantum computing solutions against established classical methods.

Methods

The core of this work lies in an enhanced mathematical model for currency arbitrage. The key innovation is the introduction of simple-cycle preservation constraints. These constraints ensure that any identified arbitrage opportunity represents a valid, non-repeating sequence of currency exchanges, preventing the inclusion of redundant or impossible trade loops. This enhanced model is formulated as a Binary Quadratic Programming (BQP) problem, which can be translated into a Quadratic Unconstrained Binary Optimization (QUBO) problem for compatibility with quantum solvers.

The authors then benchmark several approaches:

  • Quantum Annealing (QA): A hardware-based quantum approach designed for optimization problems.
  • Adaptive Cost Encoding (ACE): A gate-based quantum algorithm that significantly reduces qubit requirements for solving QUBO problems, making it suitable for noisy intermediate-scale quantum (NISQ) devices. They also introduce a classical multi-bit swap post-processing step (ACE-LS) to further refine ACE's solutions.
  • Classical Solvers:
    • Gurobi: A powerful commercial solver for linear and mixed-integer programming.
  • Tabu Search (TS): A metaheuristic local search algorithm.

Datasets

The experimental analysis uses real-world currency exchange rate data for 14 currencies at a specific timestamp, sourced from OANDA. This dataset is used to construct the mathematical model and the QUBO formulation for evaluating the different solvers.

Key Insights and Ideas

  • Simple-cycle constraints are crucial: The inclusion of simple-cycle preservation constraints is a significant improvement, ensuring the validity and uniqueness of arbitrage opportunities, which was a limitation in previous models.
  • Quantum offers speed advantages: Quantum solvers (QA and ACE-LS) demonstrate a clear advantage in execution time, scaling linearly with problem size, whereas classical solvers like Gurobi exhibit exponential growth. This speed is critical for real-time arbitrage.
  • ACE-LS shows promise: The ACE-LS approach, combining a qubit-efficient quantum algorithm with classical post-processing, provides competitive solution quality and impressive speedups, making it a strong contender for practical applications.
  • Trade-offs exist: While quantum methods excel in speed, their solution quality is comparable to, but not always superior to, classical benchmarks like Gurobi on smaller problems. However, for larger, more complex problems, the speed advantage of quantum becomes paramount.

Key Results

  • Profit Generation: QA and ACE-LS achieve comparable arbitrage profits to Gurobi and Tabu Search. For instance, in the (14, 10) problem size, ACE-LS yields a profit of 11.79 pips, very close to Gurobi's 11.79 pips.
  • Execution Time: Figure 4 vividly illustrates the speed advantage. For the largest problem size (14, 10), QA and ACE-LS take a fraction of a second, while Gurobi takes significantly longer, and its time grows exponentially with problem size.
  • ACE-LS Optimization: Experiments with different optimizers and circuit layers for ACE showed that using Circuit 2 with two layers and the Differential Evolution (DE) optimizer yielded the lowest cost across all problem sizes.

Limitations and Future Work

The authors acknowledge that real-time currency arbitrage demands execution within microseconds to milliseconds, which is currently beyond the capabilities of current quantum hardware and HPC systems. Future work will focus on:

  • Executing ACE-LS on real quantum hardware: To validate its performance in a practical setting.
  • Developing advanced hybrid classical-quantum algorithms: To further reduce execution times and improve solution quality.
  • Exploring LS in the quantum domain: To potentially reduce classical overhead.
  • Investigating neutral atom-based quantum computers: For their scalability and long coherence times, which could further enhance computational efficiency.

Summary updated

Abstract

Currency arbitrage leverages price discrepancies in currency exchange rates across different currency pairs to gain risk-free profits. It involves multiple trading, where short-lived price discrepancies require real-time, high-speed processing of vast solution space, posing challenges for classical computing. In this work, we formulate an enhanced mathematical model for the currency arbitrage problem by adding simple cycle preservation constraints, which guarantee trading cycle validity and eliminate redundant or infeasible substructures. To solve this model, we use and benchmark various solvers, including Quantum Annealing (QA), gate-based quantum approaches such as Variational Quantum Algorithm with Adaptive Cost Encoding (ACE), as well as classical solvers such as Gurobi and classical meta heuristics such as Tabu Search (TS). We propose a classical multi-bit swap post-processing to improve the solution generated by ACE. Using real-world currency exchange data, we compare these methods in terms of both arbitrage profit and execution time, the two key performance metrics. Our results give insight into the current capabilities and limitations of quantum methods for real-time financial use cases.
BibTeX
@misc{roy2025toward,
  author = {Suman Kumar Roy and Rahul Rana and M Girish Chandra and Niraj Kumar and Manoj Nambiar},
  title = {{Toward Quantum Enabled Solutions for Real-Time Currency Arbitrage in Financial Markets}},
  year = {2025},
  doi = {10.1109/aimlsystems67835.2025.11330637},
}
Figure