Execution Aware A* for Cross Exchange Stablecoin Arbitrage

Kevin Litvin
Proceedings of the The 39th Canadian Conference on Artificial Intelligence, PMLR 318:1186-1190, 2026.

Abstract

Cross-exchange cryptocurrency arbitrage enables low-risk profit from price discrepancies across exchanges, yet existing approaches employ negative cycle detection that targets opportunity identification rather than execution feasibility. We introduce an execution-aware pathfinding framework using A* search with domain-specific guidance heuristics, applied to stablecoins, a unique asset class exceeding $300 billion in market capitalization that bridges cryptocurrency and fiat currency, offering a novel dataset for arbitrage research. The problem is modelled as a weighted directed graph where nodes represent (exchange, stablecoin) pairs across 12 centralized exchanges and edges encode real-world costs including fees, slippage, gas, transfer delays, and exchange reliability. Three guidance heuristics and a multi-start strategy are evaluated over 7,200 search instances. Our slippage-aware heuristic h2 reduces node expansions by 29% relative to Dijkstra while matching its profit, demonstrating that domain-specific heuristics can meaningfully improve execution feasibility in real-time arbitrage planning. Code: https://github.com/kevinl03/Stablecoin-CrossExchange-Arbitrage.

Cite this Paper

BibTeX
@InProceedings{pmlr-v318-litvin26a, title = {Execution Aware A* for Cross Exchange Stablecoin Arbitrage}, author = {Litvin, Kevin}, booktitle = {Proceedings of the The 39th Canadian Conference on Artificial Intelligence}, pages = {1186--1190}, year = {2026}, editor = {Bouzar-Benlabiod, Lydia and Leung, Carson}, volume = {318}, series = {Proceedings of Machine Learning Research}, month = {25--29 May}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v318/main/assets/litvin26a/litvin26a.pdf}, url = {https://proceedings.mlr.press/v318/litvin26a.html}, abstract = {Cross-exchange cryptocurrency arbitrage enables low-risk profit from price discrepancies across exchanges, yet existing approaches employ negative cycle detection that targets opportunity identification rather than execution feasibility. We introduce an execution-aware pathfinding framework using A* search with domain-specific guidance heuristics, applied to stablecoins, a unique asset class exceeding $300 billion in market capitalization that bridges cryptocurrency and fiat currency, offering a novel dataset for arbitrage research. The problem is modelled as a weighted directed graph where nodes represent (exchange, stablecoin) pairs across 12 centralized exchanges and edges encode real-world costs including fees, slippage, gas, transfer delays, and exchange reliability. Three guidance heuristics and a multi-start strategy are evaluated over 7,200 search instances. Our slippage-aware heuristic h2 reduces node expansions by 29% relative to Dijkstra while matching its profit, demonstrating that domain-specific heuristics can meaningfully improve execution feasibility in real-time arbitrage planning. Code: https://github.com/kevinl03/Stablecoin-CrossExchange-Arbitrage.} }
Endnote
%0 Conference Paper %T Execution Aware A* for Cross Exchange Stablecoin Arbitrage %A Kevin Litvin %B Proceedings of the The 39th Canadian Conference on Artificial Intelligence %C Proceedings of Machine Learning Research %D 2026 %E Lydia Bouzar-Benlabiod %E Carson Leung %F pmlr-v318-litvin26a %I PMLR %P 1186--1190 %U https://proceedings.mlr.press/v318/litvin26a.html %V 318 %X Cross-exchange cryptocurrency arbitrage enables low-risk profit from price discrepancies across exchanges, yet existing approaches employ negative cycle detection that targets opportunity identification rather than execution feasibility. We introduce an execution-aware pathfinding framework using A* search with domain-specific guidance heuristics, applied to stablecoins, a unique asset class exceeding $300 billion in market capitalization that bridges cryptocurrency and fiat currency, offering a novel dataset for arbitrage research. The problem is modelled as a weighted directed graph where nodes represent (exchange, stablecoin) pairs across 12 centralized exchanges and edges encode real-world costs including fees, slippage, gas, transfer delays, and exchange reliability. Three guidance heuristics and a multi-start strategy are evaluated over 7,200 search instances. Our slippage-aware heuristic h2 reduces node expansions by 29% relative to Dijkstra while matching its profit, demonstrating that domain-specific heuristics can meaningfully improve execution feasibility in real-time arbitrage planning. Code: https://github.com/kevinl03/Stablecoin-CrossExchange-Arbitrage.
APA
Litvin, K.. (2026). Execution Aware A* for Cross Exchange Stablecoin Arbitrage. Proceedings of the The 39th Canadian Conference on Artificial Intelligence, in Proceedings of Machine Learning Research 318:1186-1190 Available from https://proceedings.mlr.press/v318/litvin26a.html.

Related Material