---
title: Portfolio Variable Selection
url: https://www.ml-quant.com/papers/ssrn/5059973/
site: ML-Quant (https://www.ml-quant.com)
updated: 2026-09-26
license: Summaries CC BY 4.0; links go to the original sources
index: https://www.ml-quant.com/llms.txt
identifier: SSRN 5059973
source_url: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5059973
featured: 2024-12-18
citations: unknown
topic: Portfolio & Allocation
---


# Portfolio Variable Selection

Machine learning methods can choose predictors for optimal portfolio choice, improving portfolio performance and decreasing portfolio risk, leading to high Sharpe ratios.

- Source: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=5059973
- Identifier: SSRN 5059973
- Released: 2024-12-16
- First featured: Quant Letter No. 79 (2024-12-18): https://www.ml-quant.com/issues/2024-12-18/
- Citations (Semantic Scholar): not tracked
- Published in: not yet
- Topic: Portfolio & Allocation

## Related

- [Cost-aware Portfolios in a Large Universe of Assets](https://www.ml-quant.com/papers/arxiv/2412.11575/): Research suggests portfolio models that include transaction costs, highlighting the importance of considering these costs when rebalancing a portfolio.
- [Does Social Pension Insurance Increase the Efficiency of Household Financial Portfolios?](https://www.ml-quant.com/papers/ssrn/5046037/): Research shows that social pension insurance in China enhances the efficiency of household financial portfolios due to changes in risk attitude and precautionary savings.
- [Smart leverage? Rethinking the role of Leveraged Exchange Traded Funds in constructing portfolios to beat a benchmark](https://www.ml-quant.com/papers/arxiv/2412.05431/): The paper investigates the potential of Leveraged Exchange Traded Funds (LETFs) in long-term investment strategies, using a neural network approach to devise strategies that beat standard benchmarks.
- [An Integral Equation in Portfolio Selection with Time-Inconsistent Preferences](https://www.ml-quant.com/papers/arxiv/2412.02446/): The article suggests a comprehensive framework for time-consistent portfolio selection, demonstrating the existence and uniqueness of a solution for the integral equation under certain conditions.
- [Stock Recommendations for Individual Investors: A Temporal Graph Network Approach with Mean-Variance Efficient Sampling](https://www.ml-quant.com/papers/doi/10-1145-3677052-3698662/): The study introduces a new model, PfoTGNRec, for stock recommendation systems that balances customer preferences with suggesting high ROI portfolios, showing superior performance on real-world individual trading data.
- [Synthetic data for portfolios: a throw of the dice will never abolish chance](https://www.ml-quant.com/papers/arxiv/2501.03993/): The paper highlights the shortcomings of generative models in finance, suggesting a method for generating multivariate returns that meets standard evaluations and aligns with observed asset returns.