{"paper_id":"2511.02136","title":"JaxMARL-HFT: GPU-Accelerated Large-Scale Multi-Agent Reinforcement Learning for High-Frequency Trading","authors":["Valentin Mohl","Sascha Frey","Reuben Leyland","Kang Li","George Nigmatulin","Mihai Cucuringu","Stefan Zohren","Jakob Foerster","Anisoara Calinescu"],"paper_date":"2025-11-03","paper_type":"software-tool","paper_url":"https://arxiv.org/pdf/2511.02136v1","page_url":"https://thequant.space/flowcharts/jaxmarl-hft-gpu-accelerated-large-scale-multi-agent-reinfor/","scores":{"empirical_rigor":8,"math_complexity":4,"quadrant":"Street Traders"},"proposes":{"status":"available","statement":"The paper proposes JaxMARL-HFT, a GPU‑accelerated open‑source multi‑agent reinforcement learning environment for high‑frequency trading on market‑by‑order data.","quote":"we introduce JaxMARL-HFT (JAX-based Multi-Agent Reinforcement Learning for High-Frequency Trading), the first GPU-accelerated open-source multi-agent reinforcement learning environment for high-frequency trading (HFT) on market-by-order (MBO) data.","summary":"Agent-based modelling (ABM) approaches for high-frequency financial markets are difficult to calibrate and validate, partly due to the large parameter space created by defining fixed agent policies. Multi-agent reinforcement learning (MARL) enables more realistic agent behaviour and reduces the…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"Order quantities are fixed a priori via an environment configuration variable.","type":"execution","stated":"explicit","quote":"The quantity associated with each order is fixed a priori through an environment configuration variable: the agents following these strategies will quote constant volumes associated with each bid and ask message."},{"assumption":"Agents receive constant volume orders equal to the minimum lot size, assumed realistic for market participants.","type":"market","stated":"explicit","quote":"In practice, this quantity can be set to the minimum lot size, which, depending on the asset price, is often the most commonly used trade size by market participants, making this assumption reasonably realistic."},{"assumption":"Reward functions are parameterised by adjustable hyper‑parameters, allowing diverse risk tolerances.","type":"model","stated":"explicit","quote":"All functions are parameterised by adjustable hyperparameters, which presents the user with a highly diverse set of possible reward functions."},{"assumption":"Posted orders from previous steps are cancelled at each RL step, which disadvantages market making in a price‑time priority LOB.","type":"execution","stated":"explicit","quote":"It is assumed that one of the hindering factors which makes profitable market making difficult is the fact that posted orders from previous steps are cancelled at each RL step."},{"assumption":"Episodes can be defined to overlap and are initialised with the LOB state at that time.","type":"model","stated":"explicit","quote":"At the beginning of each episode, the LOB is initialised as described in Frey et al. [12], using the current state of the book at that time."}]},"datasets":{"status":"available","items":[{"name":"LOBSTER AMZN order data","provider":"LOBSTER","kind":"order-book","access":"public","period":"one year","universe":"AMZN","quote":"... a year’s worth of pre-processed AMZN order data to occupy just 4GB of GPU memory","known_id":"lobster"}],"known":[{"id":"abides","name":"ABIDES simulator","provider":"ABIDES","kind":"synthetic","access":"public","quote":"We use the open-source code of ABIDES-gym and PyMarketSim to compare all implementations on the same hardware. For our experim","mentions":17,"used_in":4},{"id":"lobster","name":"LOBSTER","provider":"LOBSTER","kind":"order-book","access":"commercial","quote":"Using one day of tick-level order book data from the LOBSTER dataset [16], they show that the resulting simulation reproduces several stylised facts observed in the real data.","mentions":4,"used_in":1}],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"l2-book"},"transaction_costs":{"status":"available","modeled":"yes","headline_net_of_costs":null,"detail":"Execution cost reduced by up to 20% of a tick on average compared to TWAP.","quote":"execution cost can be decreased by up to 20% of a tick on average as compared to TWAP","source":"evidence_card","rule":"R2","candidates":{"front_matter":"unclear","evidence_card":"yes","evidence_card_quoted":true},"card_detail":"Execution cost reduced by up to 20% of a tick on average compared to TWAP.","card_quote":"execution cost can be decreased by up to 20% of a tick on average as compared to TWAP","meaning":"costs are part of the reported headline results"},"results":{"status":"available","evaluation_type":"unclear","findings":[{"finding":"Training throughput is increased by up to 240× compared with state‑of‑the‑art CPU implementations.","kind":"performance","quote":"Leveraging JAX enables up to a 240x reduction in end-to-end training time, compared with state-of-the-art reference implementations on the same hardware."},{"finding":"Learned policies outperform baseline TWAP execution and Avellaneda‑Stoikov market‑making models in their respective quality metrics.","kind":"performance","quote":"Both baselines are outperformed by the learned agent in terms of their respective quality metrics (Portfolio Value and Slippage)."},{"finding":"Market‑making agents lose about 0.2 ticks on average but this loss is considered a surprisingly good result given the environment constraints.","kind":"negative","quote":"none of the learned strategies for market making are able to make net profits, but instead lose approximately 0.2 ticks on average."},{"finding":"Execution agents achieve up to a 20% tick reduction in execution cost relative to TWAP, likely by trading more passively while guaranteeing execution.","kind":"performance","quote":"execution cost can be decreased by up to 20% of a tick on average as compared to TWAP, likely by trading more passively, whilst still guaranteeing execution."}],"reported_performance":null,"out_of_sample":{"method":"unclear","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":30,"citations":{"cited_by_in_corpus":1,"by_independent_authors":1,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":1,"unclassified":0},"independent_papers":[],"scope":"Citations between papers in this corpus only; citations from outside it are not counted."}},"implementation":{"status":"available","public":true,"code_url":"https://github.com/vmohl/JaxMARL-HFT","note":null},"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":1,"extracted_at":"2026-11-14T17:23:09Z"},"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/"}