Paper: arXiv 2503.08287
Authors: Ryan Donnelly, Zi Li
Abstract
We study a multi-agent setting in which brokers transact with an informed trader. Through a sequential Stackelberg-type game, brokers manage trading costs and adverse selection with an informed trader. In particular, supplying liquidity to the informed traders allows the brokers to speculate based on the flow information. They simultaneously attempt to minimize inventory risk and trading costs with the lit market based on the informed order flow, also known as the internalization-externalization strategy. We solve in closed form for the trading strategy that the informed trader uses with each broker and propose a system of equations which classify the equilibrium strategies of the brokers. By solving these equations numerically we may study the resulting strategies in equilibrium. Finally, we formulate a competitive game between brokers in order to determine the liquidity prices subject to precommitment supplied to the informed trader and provide a numerical example in which the resulting equilibrium is not Pareto efficient.
Complexity vs Empirical Score
- Math Complexity: 8.0/10
- Empirical Rigor: 3.0/10
- Quadrant: Lab Rats — high math complexity, limited empirical testing
Why this score: The paper presents a theoretical model with a complex Stackelberg game, PDEs, and a system of equilibrium equations that require numerical solving, indicating high math complexity. However, it focuses entirely on theoretical modeling and numerical simulation without presenting real-world backtests, datasets, or implementation details, placing it in the Lab Rats quadrant.
Research Flowchart
flowchart TD A["Research Goal:<br>Model Broker Liquidity Supply<br>to an Informed Trader"] --> B["Methodology:<br>Multi-agent Sequential<br>Stackelberg Game"] B --> C["Key Inputs:<br>Informed Trader's Private<br>Signal & Asset Payoff"] C --> D["Process 1: Closed-Form Solution<br>Informed Trader's Strategy<br>with Brokers"] D --> E["Process 2: System of Equations<br>Classifying Broker<br>Equilibrium Strategies"] E --> F["Computation:<br>Numerical Solution of<br>Broker Competitive Game"] F --> G["Key Outcomes:<br>Equilibrium Liquidity Prices<br>& Pareto Inefficiency"]
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 multi‑agent Stackelberg‑type model of brokers transacting with an informed trader, derives closed‑form informed‑trader strategies, formulates a system of equilibrium equations for broker strategies, and provides a numerical example showing the resulting equilibrium is not Pareto
Source sentence
We solve in closed form for the trading strategy that the informed trader uses with each broker and propose a system of equations which classify the equilibrium strategies of the brokers. By solving these equations numerically we may study the resulting strategies in equilibrium.
Assumes 6
- The identity of each trader is known to the brokers and quotes are customized accordingly. market
Source sentence
We assume that the identity of each trader is known to the brokers and a customized quote is made correspondingly.
- Brokers know the informed trader’s model, parameter values and initial inventory. model
Source sentence
We assume that the brokers know the informed trader’s model and the value of the model parameters, and the initial inventory of the informed trader3.
- Brokers observe the lit‑market mid‑price process but do not directly observe the alpha component. model
Source sentence
The brokers observes the mid-price process (Sν t )t∈T of the LOB in the lit market, but do not directly observe the alpha component (2).
- The alpha process is mean‑reverting with dynamics dαt = −θ αt dt + η dWαt. model
Source sentence
This additional infor-mation is generally known as ‘alpha’ and we assume that it is mean-reverting with dynamics given by equation (2).
Show 2 more
- Temporary price impact is linear in the broker’s trading speed with positive coefficient kj. execution
Source sentence
Specifically, the transaction price achieved by broker j is ˆSj t = Sν t + kj νj t , j ∈{1, . . . , N} , where kj > 0 is the temporary price impact parameter.
- Liquidity cost parameters κj and cj are positive and known to the informed trader. execution
Source sentence
The quotes (i.e., execution prices if there is a trade) for the informed and uninformed traders are ˆSI,j t = Sν t + κj ωj t and ˆSU,j t = Sν t + cj uj t , j ∈{1, . . . , N} , (3) respectively, where the liquidity cost parameters κj > 0 are known by the informed trader and cj > 0 a
Data
Synthetic data only.
Costs
Modeled
Temporary price impact kj νj t and linear liquidity costs κj ωj t (informed) and cj uj t (uninformed) are incorporated in the price equations.
Source sentence
Specifically, the transaction price achieved by broker j is ˆSj t = Sν t + kj νj t , j ∈{1, . . . , N} , where kj > 0 is the temporary price impact parameter. The quotes (i.e., execution prices if there is a trade) for the informed and uninformed traders are ˆSI,j t = Sν t + κj ωj t and ˆSU,j t = Sν t + cj uj t , j ∈{1, . . . , N} , (3) respectively, where the liquidity cost parameters κj > 0 are known by the
Finds 5
- The computed liquidity‑price Nash equilibrium for two brokers is not Pareto efficient. theoretical
Source sentence
Finally, we formulate a competitive game between brokers in order to determine the liquidity prices subject to precommitment supplied to the informed trader and provide a numerical example in which the resulting equilibrium is not Pareto efficient.
- Numerical experiments are performed for N=2 brokers, solving the equilibrium via an iterative algorithm. theoretical
Source sentence
For reasons related to computational complexity and presentability of the results, we only consider the situation of N = 2 brokers.
- Broker 1’s initial value function V1(κ1,κ2) is concave in κ1 for any fixed κ2. theoretical
Source sentence
We can observe that for fixed κ2, initial value function V 1 is concave with respect to κ1.
- Increasing the running inventory penalty ϕ1 reduces broker 1’s speculative trading volume and average inventory level. performance
Source sentence
We see that the average inventory level of broker 1 decreases from 0.17 to 0.02, which indicates broker 1’s trading behavior is indeed suppressed due to the higher running inventory penalty. These observations rationalize the decreasing tendency of κ1∗with respect to ϕ1 in Figure 5: broker 1 reduces her trading activities to avoid a higher running inventory penalty, and she prefers to decrease her speculative
- The equilibrium liquidity price κ*1 responds differently to terminal inventory penalty a1 depending on the level of running inventory penalty ϕ1. theoretical
Source sentence
On the left panel of Figure 9, we can see that the optimal liquidity price κ∗ 1 set by broker 1 increases for higher terminal penalty a1, at high running inventory penalty ϕ1. However, on the right panel, for low running inventory penalty ϕ1, the optimal cost κ∗ 1 is decreasing in terminal inventory penalty a1.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 29 related papers). Coverage is partial, so this is neither replication nor refutation.
3 papers in the archive : Strategic Learning and Trading in Broker-Mediated Markets
Code
No repository by the authors found in the paper.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.