Paper: arXiv 2412.20847

Authors: Alif Aqsha, Fayçal Drissi, Leandro Sánchez-Betancourt

Abstract

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 trader estimates the broker’s trading activity in the lit market while the broker estimates the informed trader’s private signal. Brokers hold a strategic advantage over traders who rely solely on prices to filter information. We find that information leakage in the client’s trading flow yields an economic value to the broker that is comparable to transaction costs; she speculates profitably and mitigates risk effectively, which, in turn, adversely impacts the informed trader’s performance. In contrast, low signal-to-noise sources, such as prices, result in the broker’s trading performance being indistinguishable from that of a naive strategy that internalises noise flow, externalises informed flow, and offloads inventory at a constant rate.

Complexity vs Empirical Score

  • Math Complexity: 9.0/10
  • Empirical Rigor: 2.0/10
  • Quadrant: Lab Rats — high math complexity, limited empirical testing

Why this score: The paper employs advanced stochastic control, filtering (Kalman-Bucy, Riccati equations), and game theory, indicating high mathematical complexity. However, it relies on theoretical models and simulations without mentioning real backtests or implementation details, showing low empirical rigor.

Research Flowchart

  flowchart TD
  A["Research Goal:<br>How do strategic interactions between<br>a broker and an informed trader affect<br>market outcomes and performance?"] --> B["Model & Methodology:<br>Dynamic Microstructure Model<br>Filtering & Inventory Control"]
  B --> C["Key Mechanisms:<br>1. Broker estimates trader's signal<br>2. Trader estimates broker's lit activity<br>3. Inventory risk & info leakage"]
  C --> D{"Computational Analysis:<br>Simulation & Equilibrium Solving"}
  D --> E["Performance Comparison<br>vs. Naive Strategy"]
  E --> F["Key Findings/Outcomes:<br>• Broker's info advantage creates value<br>• Profitable speculation & risk mitigation<br>• Adverse impact on informed trader<br>• Low signal-to-noise prices =<br>indistinguishable from naive strategy"]

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

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.

Source sentence
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.

Assumes 6

  • The broker provides liquidity to both the informed and uninformed traders and manages inventory by trading in a lit market. market
    Source sentence
    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.
  • Transaction costs charged by the broker are linear in the clients' trading speeds with positive coefficients b (informed) and c (uninformed). execution
    Source sentence
    As remuneration for this service, the broker charges transaction costs to both traders.
  • The broker incurs linear transaction costs a for her own trading and generates a linear permanent price impact with coefficient p>0. execution
    Source sentence
    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.
  • 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. model
    Source sentence
    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.
Show 2 more
  • 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). model
    Source sentence
    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.
  • The broker’s learning problem is formulated as a linear‑quadratic stochastic control problem leading to a matrix Riccati differential equation. model
    Source sentence
    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.

Data

Synthetic data only.

Costs

Modeled Costs are part of the reported headline results.

Linear transaction costs with coefficients a for the broker's own trades, b for the informed trader, and c for the uninformed trader.

Source sentence
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.

Finds 5

  • 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. theoretical
    Source sentence
    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.
  • 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. performance
    Source sentence
    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
  • 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. performance
    Source sentence
    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.
  • The broker increases externalisation and speculative trading when signal values are high or when its signal estimates are more precise. theoretical
    Source sentence
    In particular, the broker increases externalisation and speculative trading when signal values are high or when her estimates are more precise.
  • The best‑performing estimator remains robust to inventory misspecification during the early part of the trading horizon. theoretical
    Source sentence
    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 .

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 15 related papers). Coverage is partial, so this is neither replication nor refutation.

Cited by 5 papers in the archive (3 by independent authors)

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.

Cited References (80)

  1. Lectures on BSDEs, stochastic control, and stochastic differential games with financial applications (2016)
  2. Fundamentals of Stochastic Filtering (2009)
  3. Mean field games with partial information for algorithmic trading (2018) — link
  4. Optimal dealer pricing under transactions and return uncertainty (1981)
  5. Pathwise stochastic control with applications to robust filtering (2020)
  6. Portfolio choice, portfolio liquidation, and portfolio transition under drift uncertainty (2019)
  7. Optimal execution: A review (2022)
  8. Parameter uncertainty in the Kalman–Bucy filter (2019)
and 72 more
  1. Liquidity in competitive dealer markets (2021)
  2. Optimal execution of portfolio transactions (2001)
  3. Liquidity provision with adverse selection and inventory costs (2023)
  4. The Financial Mathematics of Market Liquidity: From optimal execution to market making (2016)
  5. Estimation and Control of Dynamical Systems (2018)
  6. New Results in Linear Filtering and Prediction Theory (1961)
  7. Algorithmic market making in dealer markets with hedging and market impact (2023)
  8. Convex analysis and variational problems (1999)
  9. Internalisation by electronic FX spot dealers (2019)
  10. A mean-field game of market-making against strategic traders (2023)
  11. Optimal trading strategy and supply/demand dynamics (2013)
  12. On the impossibility of informationally efficient markets (1980)
  13. Liquidity and market structure (1988)
  14. Detecting Toxic Flow — → 2312.05827
  15. Nash Equilibrium between Brokers and Traders — → 2407.10561
  16. Unwinding Stochastic Order Flow: When to Warehouse Trades — → 2310.14144
  17. Strategic insider trading equilibrium: a filter theory approach (2012)
  18. Liquidity competition between brokers and an informed trader — → 2503.08287
  19. Double-execution strategies using path signatures (2022)
  20. Trading foreign exchange triplets (2020)
  21. A mean field game between informed traders and a broker — → 2401.05257
  22. A survey of nonsymmetric Riccati equations (2002)
  23. Non—Blow—Up Conditions for Riccati—type Matrix Differential and Difference Equations (2000)
  24. Optimal signal-adaptive trading with temporary and transient price impact (2022)
  25. Optimal trading: The importance of being adaptive (2021)
  26. Mean-field games with differing beliefs for algorithmic trading (2020)
  27. Trading algorithms with learning in latent alpha models (2019)
  28. Merton’s optimal investment problem with jump signals (2021) — link
  29. Stochastic calculus and applications (2015)
  30. Enhancing trading strategies with order book signals (2018)
  31. Option Volume Imbalance as a predictor for equity market returns (2022) — link
  32. Continuous-Time Dynamic Decision Making with Costly Information (2024) — link
  33. Optimal adaptive control with separable drift uncertainty (2023) — link
  34. Unwinding Toxic Flow with Partial Information — → 2407.04510
  35. Optimal execution with stochastic delay (2021)
  36. Brokers and informed traders: dealing with toxic flow and extracting trading signals (2025)
  37. Liquidation in target zone models (2018)
  38. Closed-loop N ash competition for liquidity (2023)
  39. Continuous-time Stochastic Control and Optimization with Financial Applications (2009)
  40. Portfolio liquidation games with self-exciting order flow (2022)
  41. Liquidity in competitive dealer markets (2021)
  42. Investing for the long run when returns are predictable (2000)
  43. Optimal Trade Execution under Endogenous Order Flow (2023)
  44. Algorithmic trading with model uncertainty (2017)
  45. Continuous-time stochastic control and optimization with financial applications (2009)
  46. Incorporating signals into optimal trading (2019)
  47. Incorporating order-flow into optimal execution (2016)
  48. Algorithmic and high-frequency trading (2015)
  49. Buy low, sell high: A high frequency trading perspective (2014)
  50. Strategic Bonding Curves in Automated Market Makers (2024)
  51. Matrix Riccati equations in control and systems theory (2012)
  52. Fast and Slow Optimal Trading with Exogenous Information (2023)
  53. Models of market liquidity: Applications to traditional markets and automated market makers (2023)
  54. Multi-asset Optimal Execution and Statistical Arbitrage Strategies under Ornstein–Uhlenbeck Dynamics (2022)
  55. Market microstructure theory (1998)
  56. Mean-variance hedging for partially observed drift processes (2001)
  57. Solvability of differential R iccati equations and applications to algorithmic trading with signals (2022)
  58. Execution in an aggregator (2017)
  59. The dynamics of dealer markets under competition (1983)
  60. Optimal control of execution costs (1998)
  61. Optimal Portfolio in Partially Observed Stochastic Volatility Models (2001)
  62. Equilibrium in Functional Stochastic Games with Mean-Field Interaction — → 2306.05433
  63. High Frequency Trading in a Limit Order Book (2008)
  64. Informed speculation with imperfect competition (1989)
  65. Continuous auctions and insider trading (1985)
  66. Buy Low, Sell High: A High Frequency Trading Perspective (2014) — link
  67. Market making with alpha signals (2020)
  68. Riccati Differential Equations (1972)
  69. PORTFOLIO OPTIMIZATION IN A DEFAULT MODEL UNDER FULL/PARTIAL INFORMATION (2015)
  70. Optimal Market Making under Partial Information with General Intensities (2020)
  71. Mean Field Game Systems with Common and Latent Processes (2018)
  72. Broker-Trader Partial Information Nash Equilibria — → 2412.17712

8 of 80 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.