Paper: arXiv 2311.14318
Authors: Lijun Bo, Yijie Huang, Xiang Yu
Abstract
This paper studies an infinite horizon optimal tracking portfolio problem using capital injection in incomplete market models. The benchmark process is modelled by a geometric Brownian motion with zero drift driven by some unhedgeable risk. The relaxed tracking formulation is adopted where the fund account compensated by the injected capital needs to outperform the benchmark process at any time, and the goal is to minimize the cost of the discounted total capital injection. When model parameters are known, we formulate the equivalent auxiliary control problem with reflected state dynamics, for which the classical solution of the HJB equation with Neumann boundary condition is obtained explicitly. When model parameters are unknown, we introduce the exploratory formulation for the auxiliary control problem with entropy regularization and develop the continuous-time q-learning algorithm in models of reflected diffusion processes. In some illustrative numerical example, we show the satisfactory performance of the q-learning algorithm.
Complexity vs Empirical Score
- Math Complexity: 8.5/10
- Empirical Rigor: 4.0/10
- Quadrant: Lab Rats — high math complexity, limited empirical testing
Why this score: The paper presents advanced mathematical theory involving reflected diffusions, HJB equations with Neumann boundary conditions, and continuous-time q-learning extensions, resulting in high math complexity. However, the empirical component relies on a single illustrative numerical example and a comparison with MLE on real data, lacking extensive backtesting or implementation details for a trading strategy.
Research Flowchart
flowchart TD
A["Research Goal:<br>Optimal tracking portfolio in incomplete markets<br>with capital injection"] --> B["Model Setup<br>Benchmark: Geometric Brownian Motion<br>Formulation: Relaxed tracking"]
B --> C{"Model Parameters Known?"}
C -- Yes --> D["Known Model Method<br>Equivalent auxiliary control problem<br>Solve HJB with Neumann boundary condition<br>Explicit solution obtained"]
C -- No --> E["Unknown Model Method<br>Exploratory formulation with entropy regularization<br>Continuous-time Q-learning<br>Reflected diffusion processes"]
D --> F["Outcomes<br>Explicit analytical solution<br>Classical optimal control"]
E --> G["Outcomes<br>Q-learning algorithm performance<br>Satisfactory numerical results<br>Model-free tracking 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 introduces an exploratory formulation with entropy regularization and develops a continuous-time q‑learning algorithm for optimal tracking portfolios in incomplete markets.
Source sentence
When model parameters are unknown, we introduce the exploratory formulation for the auxiliary control problem with entropy regularization and develop the continuous-time q-learning algorithm in models of reflected diffusion processes.
Assumes 6
- Riskless interest rate is zero (r = 0). market
Source sentence
Assume that the riskless interest rate r = 0 that amounts to the change of
- Benchmark follows a geometric Brownian motion with zero drift driven by unhedgeable risk. model
Source sentence
The benchmark process is modelled by a geometric Brownian motion with zero drift driven by some unhedgeable risk.
- State dynamics are reflected (Neumann boundary condition). model
Source sentence
When model parameters are known, we formulate the equivalent auxiliary control problem with reflected state dynamics, for which the classical solution of the HJB equation with Neumann boundary condition is obtained explicitly.
- Fund account with injected capital must always outperform the benchmark (relaxed tracking). execution
Source sentence
The relaxed tracking formulation is adopted where the fund account compensated by the injected capital needs to outperform the benchmark process at any time.
Show 2 more
- Convergence of stochastic approximation as time step →0 and horizon →∞. statistical
Source sentence
we establish the convergence of the stochastic approximation algorithms for time discretization and horizon truncation when ∆t →0 and T →∞
- Temperature parameter set to γ = ρ/d for explicit fixed‑point solution. model
Source sentence
the temperature parameter γ = ρ/d, the map I has an explicit fixed point π∗∈Π given by
Data
- real market data market-prices unknown
Source sentence
we present the comparison between the q-learning algorithm and the maximum likelihood estimation (MLE) for some real market data to demonstrate the outperformance of our algorithm.
Costs
Not modeled
Finds 3
- The q‑learning algorithm shows satisfactory performance in illustrative numerical examples. performance
Source sentence
In some illustrative numerical example, we show the satisfactory performance of the q-learning algorithm.
- Learned parameters converge very satisfactorily toward their true values. theoretical
Source sentence
we illustrate the very satisfactory performance of the iteration convergence of learned parameters towards their true values.
- The q‑learning algorithm outperforms maximum likelihood estimation on real market data. performance
Source sentence
we present the comparison between the q-learning algorithm and the maximum likelihood estimation (MLE) for some real market data to demonstrate the outperformance of our algorithm.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 23 related papers). Coverage is partial, so this is neither replication nor refutation.
2 papers in the archive
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.