{"paper_id":"2412.20847","title":"Strategic Learning and Trading in Broker-Mediated Markets","authors":["Alif Aqsha","Fayçal Drissi","Leandro Sánchez-Betancourt"],"paper_date":"2024-12-30","paper_type":"theoretical","paper_url":"https://arxiv.org/pdf/2412.20847v1","page_url":"https://thequant.space/flowcharts/strategic-learning-and-trading-in-broker-mediated-markets/","scores":{"empirical_rigor":2,"math_complexity":9,"quadrant":"Lab Rats"},"proposes":{"status":"available","statement":"The paper proposes a broker‑mediated market model in which an informed trader and a broker filter each other's private information and solve stochastic control problems, comparing a broker that learns from price data with one that learns from the informed trader's order flow.","quote":"We study strategic interactions in a broker-mediated market in which agents learn and exploit each other's private information. 4.1 Learning equations The broker assumes a model where the informed trader’s trading flow is driven by his private signal.","summary":"We study strategic interactions in a broker-mediated market. A broker provides liquidity to an informed trader and to noise traders while managing inventory in the lit market. The broker and the informed trader maximise their trading performance while filtering each other’s private information; the…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The broker provides liquidity to both the informed and uninformed traders and manages inventory by trading in a lit market.","type":"market","stated":"explicit","quote":"The broker provides liquidity to the informed trader and to the uninformed trader by acting as coun-terparty to their trades throughout the trading window T. The informed trader and the uninformed trader carry out their trades at speeds η and ξ respectively.","support":"full"},{"assumption":"Transaction costs charged by the broker are linear in the clients' trading speeds with positive coefficients b (informed) and c (uninformed).","type":"execution","stated":"explicit","quote":"As remuneration for this service, the broker charges transaction costs to both traders.","support":"full"},{"assumption":"The broker incurs linear transaction costs a for her own trading and generates a linear permanent price impact with coefficient p>0.","type":"execution","stated":"explicit","quote":"We assume that the broker incurs linear transaction costs modelled by the positive constant a, and that her trading activity in the market generates a linear permanent impact modelled by the coefficient p > 0.","support":"full"},{"assumption":"The midprice follows dS_t = (p ν_t + α_t) dt + σ_S dW_t, where α_t is a stochastic signal privately observed by the informed trader.","type":"model","stated":"explicit","quote":"The midprice of the asset is driven by a stochastic signal α which is privately observed by the informed trader. Let (St)t∈T denote the process describing the midprice of the asset, whose dynamics are dSt = (p νt + αt) dt + σSdW S t , S0 ∈R+ , (2.1) where σS > 0 is the price volatility, and W S is a standard Brownian motion.","support":"full"},{"assumption":"The informed trader’s trading speed η_t and the uninformed trader’s speed ξ_t are deterministic and can be positive (buy) or negative (sell).","type":"model","stated":"explicit","quote":"The broker provides liquidity to the informed trader and to the uninformed trader by acting as coun-terparty to their trades throughout the trading window T. The informed trader and the uninformed trader carry out their trades at speeds η and ξ respectively. Positive values of the trading speed denote buying, and negative values denote selling.","support":"full"},{"assumption":"The broker’s learning problem is formulated as a linear‑quadratic stochastic control problem leading to a matrix Riccati differential equation.","type":"model","stated":"explicit","quote":"When the broker learns the signal from prices, we show that this RDE admits a unique solution, which in turn guarantees a unique optimal Markovian strategy. When the broker instead uses the informed trader’s order flow, the matrix RDE admits a unique solution up to a non-explosion condition.","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":"available","modeled":"yes","headline_net_of_costs":null,"detail":"Linear transaction costs with coefficients a for the broker's own trades, b for the informed trader, and c for the uninformed trader.","quote":"As remuneration for this service, the broker charges transaction costs to both traders. We assume that the broker incurs linear transaction costs modelled by the positive constant a, and that her trading activity in the market generates a linear permanent impact modelled by the coefficient p > 0.","source":"front_matter","rule":"R3","candidates":{"front_matter":"yes","evidence_card":"yes","evidence_card_quoted":true},"card_detail":"Linear transaction costs with coefficients a for the broker's own trades, b for the informed trader, and c for the uninformed trader.","card_quote":"As remuneration for this service, the broker charges transaction costs to both traders. We assume that the broker incurs linear transaction costs modelled by the positive constant a, and that her trading activity in the market generates a linear permanent impact modelled by the coefficient p > 0.","meaning":"costs are part of the reported headline results"},"results":{"status":"available","evaluation_type":"none","findings":[{"finding":"Information leakage from the client’s trading flow creates economic value for the broker comparable in magnitude to transaction costs; the broker can speculate profitably and manage risk more effectively, harming the informed trader’s performance.","kind":"theoretical","quote":"Information leakage in the client's trading flow generates economic value for the broker that is comparable in magnitude to transaction costs: the broker can speculate profitably and manage risk more effectively, which in turn adversely affects the informed trader's performance.","support":"full"},{"finding":"When the broker relies only on price information, its performance is indistinguishable from a naive strategy that internalises noise flow, externalises informed flow, and unwinds inventory at a constant rate.","kind":"performance","quote":"Consequently, the broker’s performance in this case resembles that of a na¨ıve strategy that internalises noise flow, externalises toxic flow, and unwinds inventory at a constant rate. We find that when brokers filter the stochastic drift αt from prices, the performance of the optimal strategy is indistinguishable from a naive strategy that externalises the order flow from the informed, internalises the uninformed","support":"full"},{"finding":"When the broker uses the informed trader’s order flow to estimate the signal, its performance improves substantially, with the improvement of the same order of magnitude as transaction costs.","kind":"performance","quote":"In contrast, when the broker uses the informed trader’s order flow to estimate the signal, her performance improves substantially. We show that the economic value of the improved filtering is of the same order of magnitude as transaction costs.","support":"full"},{"finding":"The broker increases externalisation and speculative trading when signal values are high or when its signal estimates are more precise.","kind":"theoretical","quote":"In particular, the broker increases externalisation and speculative trading when signal values are high or when her estimates are more precise.","support":"full"},{"finding":"The best‑performing estimator remains robust to inventory misspecification during the early part of the trading horizon.","kind":"theoretical","quote":"Finally, we study the robustness of the best-performing estimator and find that it remains robust to inventory misspecification during the early part of the trading horizon. Literature review .","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":15,"citations":{"cited_by_in_corpus":5,"by_independent_authors":3,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":2,"unclassified":3},"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/muhammadalifaqsha/broker_informed_noise_filtering_game","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":2,"extracted_at":"2026-11-14T19:53:13Z"},"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/"}