{"paper_id":"2510.26438","title":"An Impulse Control Approach to Market Making in a Hawkes LOB Market","authors":["Konark Jain","Nick Firoozye","Jonathan Kochems","Philip Treleaven"],"paper_date":"2025-10-30","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2510.26438v2","page_url":"https://thequant.space/flowcharts/an-impulse-control-approach-to-market-making-in-a-hawkes-lob/","scores":{"empirical_rigor":7.8,"math_complexity":9.2,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes a novel reinforcement learning approximation using a two‑network PPO architecture with self‑imitation learning, and also solves the HJB‑QVI via a deep Galerkin method.","quote":"we propose a novel Reinforcement Learning (RL) approximation inspired by auxiliary control formulations. Using a two-network PPO-based architecture with self-imitation learning, we demonstrate strong empirical performance with limited training, achieving Sharpe ratios above 30 in a realistic simulated LOB. In addition ","summary":"We study the optimal Market Making problem in a Limit Order Book (LOB) market simulated using a high-fidelity, mutually exciting Hawkes process. Departing from traditional Brownian-driven mid-price models, our setup captures key microstructural properties such as queue dynamics, inter-arrival…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"We allow for general kernel types in the Hawkes process, but restrict to exponential kernels for mathematically tractability.","type":"model","stated":"explicit","quote":"We allow for general kernel types in the Hawkes process, but restrict to exponential kernels for mathematically tractability."},{"assumption":"Impulses are restricted to be one of the 12 events in E.","type":"model","stated":"explicit","quote":"These impulses are restricted to be one of the 12 events in E for the purpose of this work."},{"assumption":"The market maker observes the LOB’s volumes at all price levels (level‑2 access).","type":"data","stated":"explicit","quote":"The market making agent observes the LOB’s volumes at all the price levels fully i.e. it has level 2 access to the LOB data."},{"assumption":"A transaction fee of 1 basis point is applied to terminal inventory liquidations.","type":"execution","stated":"explicit","quote":"we apply a transaction cost of 1 basis point (0.01%) on the absolute value of terminal inventory"},{"assumption":"The simulator runs for 5‑minute out‑of‑sample episodes for testing.","type":"data","stated":"implied","quote":"We test the policy networks on OOS data by running them through our simulator for 5 minutes over several episodes."},{"assumption":"The Hawkes LOB model includes a concave price impact due to self‑excitation.","type":"model","stated":"implied","quote":"The model implicitly includes a concave price impact due to self‑excitation."}]},"datasets":{"status":"available","items":[{"name":"Hawkes LOB simulator","provider":null,"kind":"synthetic","access":"unknown","period":null,"universe":null,"quote":"We study the optimal Market Making problem in a Limit Order Book (LOB) market simulated using a high‑fidelity, mutually exciting Hawkes process."},{"name":"Poisson LOB simulator","provider":null,"kind":"synthetic","access":"unknown","period":null,"universe":null,"quote":"Table 1: Out‑of‑sample Testing Results LOB Model Sharpe Ratio"}],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"synthetic"},"transaction_costs":{"status":"available","modeled":"yes","headline_net_of_costs":null,"detail":"1 basis point (0.01%) on the absolute value of terminal inventory","quote":"we apply a transaction cost of 1 basis point (0.01%) on the absolute value of terminal inventory","source":"evidence_card","rule":"R2","candidates":{"front_matter":"unclear","evidence_card":"yes","evidence_card_quoted":true},"card_detail":"1 basis point (0.01%) on the absolute value of terminal inventory","card_quote":"we apply a transaction cost of 1 basis point (0.01%) on the absolute value of terminal inventory","meaning":"costs are part of the reported headline results"},"results":{"status":"available","evaluation_type":"trading-pnl","findings":[{"finding":"The RL agent achieves an annualized Sharpe ratio of 31.54 after only 60 training episodes.","kind":"performance","quote":"the trained agent achieves an annualized Sharpe ratio of 31.54 within only 60 training episodes."},{"finding":"The DGM approach in the Hawkes‑only setting learns an economically unrealistic “pump and dump” strategy.","kind":"negative","quote":"the learned strategy converged to an economically unrealistic ”pump and dump” behavior exploiting the dynamic arbitrage conditions"},{"finding":"The RL policy provides symmetric liquidity provision at the top of the book and balances inventory.","kind":"theoretical","quote":"the RL formulation converges rapidly and consistently to a policy that resembles genuine market making: the agent symmetrically provides liquidity at the top of the book and dynamically balances inventory."},{"finding":"Self‑imitation learning accelerates convergence and reduces variance by focusing on rare high‑return trajectories.","kind":"theoretical","quote":"The integration of self‑imitation learning proved critical in accelerating convergence and reducing variance by focusing policy updates on rare but profitable trajectories."}],"reported_performance":null,"out_of_sample":{"method":"holdout","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":36,"citations":{"cited_by_in_corpus":2,"by_independent_authors":1,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":1,"unclassified":1},"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":1,"extracted_at":"2026-11-14T17:23:14Z"},"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/"}