Paper: arXiv 2401.13126
Authors: Nakul Upadhya, Alexandre Granzer-Guay
Abstract
This work discusses the benefits of constrained portfolio turnover strategies for small to medium-sized portfolios. We propose a dynamic multi-period model that aims to minimize transaction costs and maximize terminal wealth levels whilst adhering to strict portfolio turnover constraints. Our results demonstrate that using our framework in combination with a reasonable forecast, can lead to higher portfolio values and lower transaction costs on average when compared to a naive, single-period model. Such results were maintained given different problem cases, such as, trading horizon, assets under management, wealth levels, etc. In addition, the proposed model lends itself to a reformulation that makes use of the column generation algorithm which can be strategically leveraged to reduce complexity and solving times.
Complexity vs Empirical Score
- Math Complexity: 8.0/10
- Empirical Rigor: 7.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper uses advanced mixed-integer programming and a rolling horizon framework with column generation (high math), while also presenting numerical experiments with real-world data and providing reproducible code for backtesting (high empirical rigor).
Research Flowchart
flowchart TD A["Research Goal: Optimize Transition<br>for Small-Medium Portfolios"] --> B["Methodology: Dynamic Multi-Period<br>Model with Turnover Constraints"] B --> C["Inputs: Market Data,<br>Wealth Levels, Trading Horizons"] C --> D["Computation: Column Generation<br>Algorithm for Efficiency"] D --> E["Comparison: vs. Naive<br>Single-Period Model"] E --> F["Outcome: Higher Terminal<br>Wealth & Lower Costs"] F --> G["Conclusion: Robust across<br>various market conditions"]
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 dynamic multi‑period mixed‑integer model and two integer‑programming policies for transitioning small to medium portfolios while minimizing transaction costs and maximizing terminal wealth.
Source sentence
We propose a dynamic multi-period model that aims to minimize transaction costs and maximize terminal wealth levels whilst adhering to strict portfolio turnover constraints.
Assumes 5
- The forecast used in the receding‑horizon framework has reasonable error. model
Source sentence
One shortcoming of this work is the assumptions that the forecast will have reasonable error.
- No borrowing of cash is allowed; all cash balances must remain positive. execution
Source sentence
the transition must be made without borrowing cash (all C values must be positive) or short-selling (portfolios must be non-negative…
- Each trade incurs a fixed trading fee F. execution
Source sentence
each time we buy or sell any asset, we must pay a trading fee F.
- Retail investors have wealth below $500,000, making standard convex models inappropriate. market
Source sentence
they start breaking down when wealth levels start decreasing to “young” or “novice” investor levels (below $500 000).
Show 1 more
- The forecast model (N‑HiTS) is trained on five years of historical price and interest‑rate data. data
Source sentence
its own instance of the model which was trained for 125 epochs using 5 years of historical data from the start of the training period
Data
- Real‑world financial data and scenarios market-prices unknown
Source sentence
We demonstrate the effectiveness of our proposed method through a set of numerical experiments using real‑world financial data and scenarios.
- N‑HiTS forecast training data market-prices unknown
Source sentence
its own instance of the model which was trained for 125 epochs using 5 years of historical data from the start of the training period
Costs
Modeled
A fixed trading fee F is charged for each buy or sell transaction.
Source sentence
each time we buy or sell any asset, we must pay a trading fee F.
Finds 5
- The directional trading policy reduces the number of trades and the average money wasted on trading fees compared with a naive single‑period baseline. performance
Source sentence
directional formulation significantly reduced the number of trades needed to achieve our target portfolio compared to our naive baseline and as a result the average amount of money wasted on trading fees is also significantly lower.
- The directional trading policy also reduces losses to portfolio value in bear markets relative to a single‑period transition. performance
Source sentence
Experimental results show that the directional trading policy significantly reduces the amount of money wasted on trading fees and reduces losses to portfolio value in bear markets when compared to a single‑period transition.
- The directional penalty policy can achieve significant returns in bull markets for investors willing to accept higher risk. performance
Source sentence
The directional penalty policy also shows significant potential and allows the user to achieve significant returns in bull markets when users are willing to accept higher risk
- All proposed policies have relatively low computational load, suitable for low‑cost cloud deployment. performance
Source sentence
Additionally the computational load of all our policies is relatively low, allowing our methodologies to be easily implemented in practice, potentially on low‑cost cloud computing platforms.
- Increasing the penalty parameter λ reduces the number of trades, providing a controllable risk‑aggressiveness trade‑off. performance
Source sentence
One can also see that as the penalty parameter is increased, the number of trades reduces as well, reinforcing λ as a value to control accepted risk and aggressiveness.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 25 related papers). Coverage is partial, so this is neither replication nor refutation.
Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.
Cited References (19)
- Darts: User-Friendly Modern Machine Learning for Time Series (2022) — link
- N-HiTS: Neural Hierarchical Interpolation for Time Series Forecasting (2022) — link
- Multi- Period Trading via Convex Optimization (2017) — link
- Deep learning for forecasting stock returns in the cross-section (2018)
- Survey of stock market prediction using machine learning approach (2017) — link
- Systematic analysis and review of stock market prediction techniques (2019) — link
- Gurobi Optimizer Reference Manual (2023) — link
- Organizational Commitment Profiles and Job Satisfaction among Greek Private and Public Sector Employees (2007) — link
and 11 more
- Portfolio optimization with linear and fixed transaction costs (2007) — link
- Portfolio Selection (1952) — link
- CAPITAL ASSET PRICES : A THEORY OF MARKET EQUILIBRIUM UNDER CONDITIONS OF RISK * (1964) — link
- Linear and Mixed Integer Programming for Portfolio Optimization (2015) — link
- Optimization of conditional value-at-risk (2000) — link
- Breaking Down the Direct Indexing Investment Process (2022) — link
- Liquidity regimes and optimal dynamic asset allocation (2020) — link
- Dynamic Trading with Predictable Returns and Transaction Costs : Dynamic Trading with Predictable Returns and Transaction Costs (2013) — link
- A Transition Model for Portfolio Revision (1967) — link
- PyEPO: A PyTorch-based End-to-End Predict-then-Optimize Library for Linear and Integer Programming (2022) — link
- Smart “predict, then optimize” (2022)
0 of 19 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.