{"paper_id":"2502.15742","title":"Currency Arbitrage Optimization using Quantum Annealing, QAOA and Constraint Mapping","authors":["Sangram Deshpande","Elin Ranjan Das","Frank Mueller"],"paper_date":"2025-02-08","paper_type":"methodological","paper_url":"https://arxiv.org/pdf/2502.15742v1","page_url":"https://thequant.space/flowcharts/currency-arbitrage-optimization-using-quantum-annealing-qao/","scores":{"empirical_rigor":3,"math_complexity":8.5,"quadrant":"Lab Rats"},"proposes":{"status":"available","statement":"The paper proposes a quantum‑based approach to currency arbitrage that includes implementing the Quantum Approximate Optimization Algorithm (QAOA) with Qiskit and an N‑choose‑K methodology using D‑Wave's Ocean suite to find optimal arbitrage cycles.","quote":"In this study, we implement the Quantum Approximate Optimization Algorithm (QAOA) utilizing Qiskit version 1.2. Furthermore, we present an NchooseK-based methodology utilizing D-Wave's Ocean suite.","summary":"Currency arbitrage capitalizes on price discrepancies in currency exchange rates between markets to produce profits with minimal risk. By employing a combinatorial optimization problem, one can ascertain optimal paths within directed graphs, thereby facilitating the efficient identification of…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"A profitable arbitrage cycle is identified only if the calculated profit exceeds 1; otherwise no arbitrage path is reported.","type":"model","stated":"explicit","quote":"If the calculated profit exceeds 1, the arbitrage path is printed; otherwise, the output indicates that there is no optimal arbitrage path.","support":"full"},{"assumption":"Exchange rates are taken as given in static tables (four, five and six currency tables) and are assumed accurate for the optimization.","type":"data","stated":"explicit","quote":"TABLE I FOUR CURRENCY EXCHANGE RATES EUR USD CHF JPY EUR 1.0 1.13217 1.11777 120.756 USD 1/1.13403 1.0 0.98804 106.034 CHF 1/1.12005 1/0.99250 1.0 105.564 JPY 1/120.887 1/106.266 1/108.042 1.0 A sample currency exchange rate table from [2] is utilized, encompassing four distinct currencies (see Table I). We additionally utilize tables for five and six currency ex-change rates in Tables II and III, respectively.","support":"full"},{"assumption":"The logarithm of the product of exchange rates can be optimized linearly, allowing a linear cost function to represent arbitrage profit.","type":"model","stated":"explicit","quote":"By optimizing the logarithm of the product, the order can be reduced to linear. We proceed to introduce a linear cost function [4], C = X i,j −log(rij)bij, (1) and a penalty function for cyclic constraints [1], P = X i X j̸=j′ bi,jbi,j′ + X j X i̸=i′ bi,jbi′,j.","support":"full"},{"assumption":"Penalty weights (mp) can be chosen to enforce the one‑incoming‑one‑outgoing edge constraint for a closed cycle.","type":"model","stated":"explicit","quote":"The total cost function, Ctot, is a linear combination of C and P, namely Ctot = C + mpP, (3) where mp is a constraint factor. Various values of mp were employed, and an appropriate value was selected for inclusion in the objective function.","support":"full"},{"assumption":"The QUBO formulation can be embedded onto D‑Wave hardware using optimal embedding techniques to map qubits effectively.","type":"execution","stated":"explicit","quote":"Implementing the QUBO on D-Wave’s hardware required the application of optimal embedding techniques to effectively map our problem onto the machine’s qubit connec-tivity, thereby ensuring minimal error rates and maximizing solution accuracy.","support":"full"},{"assumption":"The COBYLA optimizer will converge to suitable QAOA parameters within a reasonable number of iterations.","type":"execution","stated":"explicit","quote":"For classical optimization of parameters within the QAOA circuit, we employed the COBYLA optimizer of the SciPy framework. In this case, the optimizer required 92 iterations to converge and determine the optimal parameters for the given objective and constraints.","support":"full"}]},"datasets":{"status":"available","items":[{"name":"Four‑currency exchange rate table","provider":null,"kind":"market-prices","access":"unknown","period":null,"universe":"EUR, USD, CHF, JPY","quote":"TABLE I FOUR CURRENCY EXCHANGE RATES EUR USD CHF JPY EUR 1.0 1.13217 1.11777 120.756 USD 1/1.13403 1.0 0.98804 106.034 CHF 1/1.12005 1/0.99250 1.0 105.564 JPY 1/120.887 1/106.266 1/108.042 1.0 A sample currency exchange rate table from [2] is utilized, encompassing four distinct currencies (see Table I).","support":"full"},{"name":"Five‑currency exchange rate table","provider":null,"kind":"market-prices","access":"unknown","period":null,"universe":"five currencies (unspecified)","quote":"We additionally utilize tables for five and six currency ex-change rates in Tables II and III, respectively.","support":"full"},{"name":"Six‑currency exchange rate table","provider":null,"kind":"market-prices","access":"unknown","period":null,"universe":"six currencies (unspecified)","quote":"We additionally utilize tables for five and six currency ex-change rates in Tables II and III, respectively.","support":"full"}],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"synthetic"},"transaction_costs":{"status":"not_applicable","modeled":"n/a","headline_net_of_costs":null,"detail":null,"quote":null,"source":"front_matter","rule":"R1","candidates":{"front_matter":"n/a","evidence_card":"no","evidence_card_quoted":false},"meaning":"the paper reports no trading or execution result"},"results":{"status":"available","evaluation_type":"none","findings":[{"finding":"The QAOA solution obtained with Qiskit for four currencies suggested trading only EUR‑USD with a profit of 0.9983598317504827, which is suboptimal (a net loss).","kind":"negative","quote":"The best solution provided by QAOA on Qiskit for four currencies suggested trading between only two currency pairs, EUR and USD, with the profit = 0.9983598317504827, which is suboptimal, i.e., a net loss.","support":"full"},{"finding":"D‑Wave quantum annealing using the N‑choose‑K Ocean Solver yielded the most efficient solution in the least amount of time.","kind":"performance","quote":"FUTURE WORKS In this study, we illustrate that the formulation of a QUBO and its resolution using D-Wave Quantum Annealers yielded the most efficient solution in the least amount of time.","support":"full"},{"finding":"The COBYLA optimizer required 92 iterations to converge to the optimal QAOA parameters for the given objective and constraints.","kind":"performance","quote":"In this case, the optimizer required 92 iterations to converge and determine the optimal parameters for the given objective and constraints.","support":"full"},{"finding":"The optimized QAOA objective value achieved was -29.242109307796706.","kind":"performance","quote":"The objective was to minimize the cost function, and the optimized value achieved was -29.242109307796706.","support":"full"},{"finding":"When running QAOA on the AerSimulator in noiseless settings, suboptimal results were observed.","kind":"negative","quote":"However, when running on the AerSim-ulator in noiseless settings, we observed suboptimal results.","support":"full"}],"reported_performance":null,"out_of_sample":{"method":"none","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":18,"citations":{"cited_by_in_corpus":0,"by_independent_authors":0,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":0,"unclassified":0},"independent_papers":[],"scope":"Citations between papers in this corpus only; citations from outside it are not counted."}},"implementation":{"status":"available","public":false,"code_url":null,"note":"No repository presented by the authors as their own was found in the paper. Absence here is not proof that no code exists."},"replication":{"status":"none_found","card":null,"note":"No replication of this paper has been run by thequant.space yet."},"extraction":{"text_source":"fulltext","model":"openai/gpt-oss-120b","version":2,"extracted_at":"2026-11-14T19:52:31Z"},"provenance":{"claims":"Everything here is what the authors state in the paper. Nothing was re-run or reproduced by the platform.","quotes":"Assumptions, datasets, cost detail and findings were extracted by a language model. Every quote is verbatim paper text. Where an item has `support`, \"full\" means every figure in it appears in the quoted sentences and \"partial\" means a figure is stated elsewhere in the paper; items with a figure found nowhere in the paper were discarded.","datasets_known":"datasets.known lists well-known vendors and public datasets by a stable id, found by a dictionary over the full text where the sentence says the data is used; it is independent of the model and covers only names in the dictionary.","flags":"transaction_costs.modeled, results.evaluation_type and results.out_of_sample.method are platform classifications (structured-decision models); see https://thequant.space/score-guide/#penalty-flags","comparisons":"supports / contradicts / qualifies are a language model's reading of two papers' quoted findings, restricted to closely related papers that share no author. They are pointers to read, not verified replications, and coverage is partial.","implementation":"code_url is a repository the paper itself presents as its own (rule-scored, ambiguous cases judged by a model), checked for liveness when it was recorded."},"docs":"https://thequant.space/agents/"}