Paper: arXiv 2501.04646

Authors: Riccardo De Blasis, Luca Galati, Filippo Petroni

Abstract

Understanding the dependencies among financial assets is critical for portfolio optimization. Traditional approaches based on correlation networks often fail to capture the nonlinear and directional relationships that exist in financial markets. In this study, we construct directed and weighted financial networks using the Mixture Transition Distribution (MTD) model, offering a richer representation of asset interdependencies. We apply local assortativity measures–metrics that evaluate how assets connect based on similarities or differences–to guide portfolio selection and allocation. Using data from the Dow Jones 30, Euro Stoxx 50, and FTSE 100 indices constituents, we show that portfolios optimized with network-based assortativity measures consistently outperform the classical mean-variance framework. Notably, modalities in which assets with differing characteristics connect enhance diversification and improve Sharpe ratios. The directed nature of MTD-based networks effectively captures complex relationships, yielding portfolios with superior risk-adjusted returns. Our findings highlight the utility of network-based methodologies in financial decision-making, demonstrating their ability to refine portfolio optimization strategies. This work thus underscores the potential of leveraging advanced financial networks to achieve enhanced performance, offering valuable insights for practitioners and setting a foundation for future research.

Complexity vs Empirical Score

  • Math Complexity: 7.5/10
  • Empirical Rigor: 6.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper employs advanced mathematics including the Mixture Transition Distribution (MTD) model, multivariate Markov chains, and network assortativity metrics, which require significant theoretical understanding. It is also empirically rigorous, using real-world data from three major indices, conducting out-of-sample backtests, and reporting performance metrics like Sharpe ratios against a benchmark.

Research Flowchart

  flowchart TD
  A["Research Goal: Improve Portfolio Optimization<br>using Directed Financial Networks"] --> B["Data: Dow Jones 30, Euro Stoxx 50, FTSE 100"]
  B --> C["Method: Construct Mixture Transition Distribution<br>(MTD) Directed Networks"]
  C --> D["Compute Local Assortativity Measures<br>on MTD Networks"]
  D --> E{"Optimization Process"}
  E --> F["Network-Based Portfolio<br>High Assortativity Allocation"]
  E --> G["Classical Mean-Variance<br>Framework Baseline"]
  F --> H["Outcomes: Network Portfolios Outperform<br>Higher Sharpe Ratios & Enhanced Diversification"]
  G --> H

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 constructing directed weighted financial networks with the Mixture Transition Distribution model and using local assortativity measures to guide portfolio selection and allocation.

Source sentence
we construct directed and weighted financial networks using the Mixture Transition Distribution (MTD) model, offering a richer representation of asset inter‑dependencies. We apply local assortativity measures—metrics that evaluate how assets connect based on similarities or differences—to guide portfolio selection and

Assumes 5

  • The multivariate sequence of asset returns satisfies a multivariate Markov property. model
    Source sentence
    (1) states that the probability of being in state k at time t+1 for the j‑th series depends only on the state h(1)_t , ..., h(n)_t occupied by all series at time t.
  • Only long positions are allowed in the portfolio (xi ≥ 0). execution
    Source sentence
    (18) only allows for long positions in the portfolio
  • A zero risk‑free rate is used for Sharpe ratio calculation. market
    Source sentence
    using a constant risk‑free rate of return equal to zero for convenience
  • Weekends and public holidays are excluded from the price series. data
    Source sentence
    public holidays have been excluded from the analysis to ensure consistency in trading days
Show 1 more
  • Portfolio performance is evaluated with a rolling 90‑day in‑sample and 30‑day out‑of‑sample window, re‑balancing every 30 trading days. execution
    Source sentence
    we use a 90‑day in‑sample window and a 30‑day out‑of‑sample period, with re‑balancing occurring every 30 trading days

Data

Backtest sample: Dow Jones 30 constituents · equities · daily

  • Daily closing prices of Dow Jones 30, Euro Stoxx 50 and FTSE 100 constituents · Datastream (Refinitiv, LSEG) market-prices commercial
    December 2004 to December 2024 · 28 DJ30 stocks, 46 STOXX50 stocks, 78 FTSE100 stocks
    Source sentence
    This study uses daily closing price data sourced from Datastream, a product of Refinitiv, an LSEG business. The dataset encompasses stocks included in three major indices: the Dow Jones 30 (DJ30), Euro Stoxx 50 (STOXX50), and FTSE 100 (FTSE100)

Recognised sources: Refinitiv / LSEG (Datastream, Eikon, Tick History)

Costs

Not modeled A trading or execution result is reported without costs.

Finds 5

Reported performance (Local assortativity‑based portfolio optimization): Sharpe 1.26 (benchmark 0.94)

  • Portfolios optimized with network‑based local assortativity measures consistently outperform the classical mean‑variance framework. performance
    Source sentence
    we show that portfolios optimized with network‑based assortativity measures consistently outperform the classical mean‑variance framework
  • The Extended Piraveenan et al. (2010) measure achieves the highest Sharpe ratio of 1.187 under the in‑out modality with Max Quadratic Utility, exceeding the Markowitz benchmark. performance
    Source sentence
    The Extended Piraveenan et al. (2010) measure achieves the highest Sharpe ratio of 1.187 under the in‑out modality with Max Quadratic Utility, exceeding the benchmark by a significant margin.
  • The out‑out modality under Max Sharpe optimization delivers a Sharpe ratio of 1.117, indicating strong performance of highly influential nodes connecting with similarly influential nodes. performance
    Source sentence
    Similarly, the out‑out modality under Max Sharpe optimization delivers a Sharpe ratio of 1.117, indicating that highly influential nodes connecting with similarly influential nodes enhance portfolio returns.
  • Using the in‑out modality, the Extended Piraveenan measure yields the highest expected return of 0.121 under Max Quadratic Utility, slightly above the Markowitz benchmark of 0.103 while maintaining comparable volatility. performance
    Source sentence
    the Extended Piraveenan et al. (2010) measure under the in‑out modality achieves the highest expected return of 0.121 under Max Quadratic Utility, slightly exceeding the Markowitz benchmark of 0.103, while maintaining a comparable annual volatility of 0.215.
  • The Peel et al. (2018) measure reaches a Sharpe ratio of 1.222 under Max Sharpe optimization for the out‑in modality, underscoring the benefit of diverse network connections. performance
    Source sentence
    1.222 under Max Sharpe optimization, underscoring the role of diverse network connections in optimizing risk‑adjusted returns.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 35 related papers). Coverage is partial, so this is neither replication nor refutation.

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.

Cited References (41)

  1. Higher Order Assortativity in Complex Networks (2017) — link
  2. Extending Assortativity: An Application to Weighted Social Networks (2021) — link
  3. Econometric Measures of Connectedness and Systemic Risk in the Finance and Insurance Sectors (2012) — link
  4. A Portfolio Construction Model Based on Sector Analysis Using Dempster-Shafer Evidence Theory and Granger Causal Network: An Application to National Stock Exchange of India (2023) — link
  5. TotalRank : Ranking without Damping (2005) — link
  6. A Deeper Investigation of PageRank as a Function of the Damping Factor (2007) — link
  7. A Network Approach to Risk Theory and Portfolio Selection (2017) — link
  8. The construction of multilayer stock network model (2021) — link
and 33 more
  1. A Multivariate Markov Chain Model for Categorical Data Sequences and Its Applications in Demand Predictions (2002) — link
  2. Asset Allocation: New Evidence through Network Approaches (2021) — link
  3. Smart Network Based Portfolios (2022) — link
  4. The Mixture Transition Distribution Approach to Networks: Evidence from Stock Markets (2023) — link
  5. On the network topology of variance decompositions: Measuring the connectedness of financial firms (2014) — link
  6. Information flow in the FTX bankruptcy: A network approach — → 2407.12683
  7. Market behaviors around bankruptcy and frozen funds withdrawal: Trading stranded assets on FTX (2024) — link
  8. Multi-likelihood methods for developing relationship networks using stock market data (2022) — link
  9. Stock market networks: The dynamic conditional correlation approach (2012) — link
  10. Hierarchical Structure in Financial Markets (1999) — link
  11. Modern portfolio theory (1952)
  12. Network Filtering for Big Data : Triangulated Maximally Filtered Graph (2017) — link
  13. Mixing Patterns in Networks (2003) — link
  14. Asset Trees and Asset Graphs in Financial Markets (2003) — link
  15. Centrality- Based Equal Risk Contribution Portfolio (2024) — link
  16. Multiscale Mixing Patterns in Networks (2018) — link
  17. A Network Approach to Portfolio Selection (2016) — link
  18. Assortative Mixing in Weighted Directed Networks (2022) — link
  19. Local Assortativeness in Scale-Free Networks (2008) — link
  20. Local Assortativeness in Scale-Free Networks (2010) — link
  21. A Model for High-Order Markov Chains (1985) — link
  22. Portfolio Optimization through a Network Approach: Network Assortative Mixing and Portfolio Diversification (2024) — link
  23. Local Assortativity in Weighted and Directed Complex Networks (2023) — link
  24. Construction and robustness of directed-weighted financial stock networks via meso-scales (2022) — link
  25. Node Assortativity in Complex Networks : An Alternative Approach (2014) — link
  26. A Network Perspective of the Stock Market (2010) — link
  27. A Tool for Filtering Information in Complex Systems (2005) — link
  28. Enhancing Portfolio Allocation : A Random Matrix Theory Perspective (2024) — link
  29. Collective Dynamics of `Small-World’ Networks (1998) — link
  30. A Network View of Portfolio Optimization Using Fundamental Information (2021) — link
  31. Cointegration Analysis and Influence Rank— A Network Approach to Global Stock Markets (2014) — link
  32. Sovereign default network and currency risk premia (2023) — link
  33. Assortativity Measures for Weighted and Directed Networks (2021) — link

1 of 41 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.