Paper: arXiv 2407.05912
Authors: Maysam Khodayari Gharanchaei, Prabhu Prasad Panda
Abstract
This paper focuses on the application of quantitative portfolio management by using integer programming and clustering techniques. Investors seek to gain the highest profits and lowest risk in capital markets. A data-oriented analysis of US stock universe is used to provide portfolio managers a device to track different Exchange Traded Funds. As an example, reconstructing of NASDAQ 100 index fund is presented.
Complexity vs Empirical Score
- Math Complexity: 6.5/10
- Empirical Rigor: 6.0/10 (capped from 7.0: transaction costs not modelled; see penalty flags)
- Quadrant: Holy Grail — high math complexity, high empirical rigor (effective empirical rigor capped from 7.0)
Why this score: The paper employs substantial mathematical formulations including integer programming, Lagrangian optimization, and quadratic variance minimization. It demonstrates strong empirical rigor by backtesting on real-world data (Yahoo Finance, WRDS Compustat) with specific train-test splits, tracking error calculations, and turnover metrics across multiple rebalancing frequencies.
Research Flowchart
flowchart TD
A["Research Goal: <br>Quantitative Portfolio Optimization"] --> B["Input: US Stock Data &<br>NASDAQ 100 Index"]
B --> C{"Methodology"}
C --> D["Clustering Algorithm"]
C --> E["Integer Programming"]
D --> F
subgraph F ["Computational Process"]
direction LR
F1["Stock Grouping"] --> F2["Selection Logic"] --> F3["Optimization"]
end
E --> F
F --> G["Key Outcomes:<br>Reconstructed Index Fund"]
G --> H["Conclusion:<br>Effective ETF Tracking Device"]
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 using integer programming combined with stock clustering to construct an index fund that tracks benchmark portfolios.
Source sentence
First of all, we used integer programming to formulate the stock clustering. This is a crucial step in constructing an index fund.
Assumes 6
- The number of selected stocks k is substantially smaller than the total number of stocks n. model
Source sentence
The most common and practical approach is to choose 𝑘 stocks, where 𝑘 is substantially smaller than 𝑛.
- Weights of selected stocks are proportional to the total market capitalization of their clusters. model
Source sentence
The weights for the selected stocks will be proportionate to total market capitalization for the cluster represented by that stock normalized the market capitalization of the 81 stocks.
- Data are split into 65% training and 35% testing for model calibration. data
Source sentence
We split the data to 65% training and 35% testing. The training data set was used to calibrate the model.
- Rebalancing can be performed quarterly, semi‑annually, or annually. execution
Source sentence
We tried three rebalancing frequencies such as quarterly, semi annually, and annually.
Show 2 more
- Daily returns are sufficient to calibrate the model. data
Source sentence
For each stock, we computed the daily returns to calibrate our model.
- Market capitalization data from WRDS Compustat are accurate and representative. data
Source sentence
In order to calculate the market capitalization, we downloaded the quarterly number of shares outstanding from ‘’WRDS Compustat’’.
Data
- NASDAQ 100 tickers benchmark public
Source sentence
We obtained the tickers for NASDAQ 100 through the ‘’pytickersymbols’’ package
- Daily closing prices market-prices public
Source sentence
then obtained daily closing prices of the tickers by using the Yahoo Finance package. The data collected ranges from 2014 to 2023.
- Quarterly shares outstanding fundamentals commercial
Source sentence
In order to calculate the market capitalization, we downloaded the quarterly number of shares outstanding from ‘’WRDS Compustat’’.
Compustat WRDS Yahoo Finance
Costs
Not modeled
Source sentence
We measure our turnover (Qian, 2005) and tracking error for each approach
Finds 3
- Using a larger number of selected stocks (higher k) leads to lower tracking error and lower turnover across all benchmarks. performance
Source sentence
We discovered that with a large value of k, we tend to get smaller tracking errors as well as turnover for all the benchmark values we tried out.
- More frequent rebalancing (quarterly vs semi‑annual vs annual) generally reduces tracking error and turnover. performance
Source sentence
We can also determine that rebalancing our portfolio more frequently generally leads to lower tracking error as well as turnover measure.
- The optimization approach yields a much lower tracking error than the heuristic approach. performance
Source sentence
Furthermore, the optimization approach yields a much lower tracking error since we have access to much more advanced tools than the heuristic approach.
Checked by others
Supported by
- Regret-Driven Portfolios: LLM-Guided Smart Clustering for Optimal Allocation
Compare the findings
This paper: More frequent rebalancing (quarterly vs semi‑annual vs annual) generally reduces tracking error and turnover.
That paper: 71% of configurations that outperformed SPY used a quarterly rebalancing schedule, indicating medium‑term rebalancing is beneficial.
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.