{"paper_id":"2307.01814","title":"Option Market Making via Reinforcement Learning","authors":["Zhou Fang","Haiqing Xu"],"paper_date":"2023-07-04","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2307.01814v2","page_url":"https://thequant.space/flowcharts/option-market-making-via-reinforcement-learning/","scores":{"empirical_rigor":3,"math_complexity":8.5,"quadrant":"Lab Rats"},"proposes":{"status":"available","statement":"The paper proposes a novel approach that combines a stochastic policy with reinforcement‑learning‑inspired techniques to determine the optimal bid‑ask spread policy for an options market maker trading multiple strikes and maturities.","quote":"In this paper, we propose a novel approach that combines a stochastic policy and reinforcement learning-inspired techniques to determine the optimal policy for posting bid-ask spreads for an options market maker who trades options with different maturities and strikes.","summary":"Market making of options with different maturities and strikes is a challenging problem due to its highly dimensional nature. In this paper, we propose a novel approach that combines a stochastic policy and reinforcement learning-inspired techniques to determine the optimal policy for posting…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The underlying asset price follows a geometric Brownian motion with deterministic (non‑stochastic) volatility σt.","type":"model","stated":"explicit","quote":"We assume that this asset follows a geometric Brownian motion with non-stochastic coefficients over the interval [0, T]. Note that we assume the volatility process σt of St to be non-stochastic.","support":"full"},{"assumption":"The equilibrium mid‑price of each option i is given by Pit = pit(St, Ki, Ti) + ξit, where pit is the expected price conditional on St.","type":"model","stated":"explicit","quote":"Furthermore, we assume the equilibrium mid-price of option i, denoted by Pit ∈R+, is expressed as follows: Pit = pit(St, Ki, Ti) + ξit, for t ∈[0, T].","support":"full"},{"assumption":"The trading horizon T is shorter than the earliest option maturity (T < min{T1,…,Tn}).","type":"market","stated":"explicit","quote":"Throughout this discussion, we consider the market maker’s decisions within the time period [0, T], where T < min{T1, · · · , Tn}.","support":"full"},{"assumption":"The underlying asset market is complete, so the terminal position ΔT incurs no liquidation cost.","type":"execution","stated":"explicit","quote":"Because of the complete market assumption for the underlying asset, we assume the position ∆T incurs no liquidation cost.","support":"full"},{"assumption":"The liquidation cost at terminal time is modeled as ψ0 × ∑i (PiT QiT)^2 with ψ0 > 0.","type":"other","stated":"explicit","quote":"To simplify our analysis, we assume: ψ(QT , ∆T , PT , ST ) = ψ0 × n X i=1 (PiT QiT )2. 4 Here, ψ0 > 0 is a positive coefficient that quantifies the impact of inventory on the market-maker’s final value.","support":"full"},{"assumption":"The correlation matrix Σt of the option price shocks ξt is deterministic, symmetric and positive‑semidefinite.","type":"statistical","stated":"explicit","quote":"Specifically, we assume the dynamics of ξt are given by dξt = ΣtdW t, where Σt ∈Rn×n is a deterministic process of symmetric positive semidefinite matrices, and W t ∈Rn is a standard n-dimensional vector of Brownian motions.","support":"full"}]},"datasets":{"status":"none_found","items":[],"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":"R4a","candidates":{"front_matter":"n/a","evidence_card":"yes","evidence_card_quoted":true},"card_detail":"A terminal liquidation cost ψ0 × ∑i (PiT QiT)^2 is included in the objective, representing inventory‑related costs.","card_quote":"To simplify our analysis, we assume: ψ(QT , ∆T , PT , ST ) = ψ0 × n X i=1 (PiT QiT )2. 4 Here, ψ0 > 0 is a positive coefficient that quantifies the impact of inventory on the market-maker’s final value.","meaning":"the paper reports no trading or execution result; the cost the extractor found is a model parameter"},"results":{"status":"available","evaluation_type":"none","findings":[{"finding":"Increasing inventory leads to a widening of the ask spread and a narrowing of the bid spread, reflecting strategic price adjustments to reduce inventory.","kind":"theoretical","quote":"A general pattern across all trading periods is that an **increase in inventory leads to a widening of the ask spread and a narrowing of the bid spread**. These adjustments make it **more attractive to buy from the market and less attractive to sell**, thus facilitating inventory offsetting in line with optimal market-making behavior.","support":"full"},{"finding":"The magnitude of bid‑ask spread adjustments is larger as the trading period approaches its end.","kind":"theoretical","quote":"In other words, as inventory increases, the bid-ask spread curves for t = 60 exhibit steeper changes compared to those at t = 0.","support":"full"},{"finding":"Options that are further out‑of‑the‑money exhibit wider bid‑ask spreads.","kind":"theoretical","quote":"A key observation is that options with strikes further from the current price (out-of-the-money) tend to exhibit wider bid-ask spreads.","support":"full"},{"finding":"Changing the inventory of one option affects not only its own spread but also the spread of another option, highlighting cross‑option inventory risk.","kind":"theoretical","quote":"Interestingly, increasing the inventory of the first option not only affects its own bid-ask spread but also influences the spread of the second option. This interdependence underscores the role of inventory risk across multiple options, highlighting the necessity of joint optimization in a multi-asset market-making strategy.","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":34,"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":0,"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":2,"extracted_at":"2026-11-14T20:02:15Z"},"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/"}