Paper: arXiv 2503.15534

Authors: Qian Hui, Tiandong Wang

Abstract

The failure of key financial institutions may accelerate risk contagion due to their interconnections within the system. In this paper, we propose a robust portfolio strategy to mitigate systemic risks during extreme events. We use the stock returns of key financial institutions as an indicator of their performance, apply extreme value theory to assess the extremal dependence among stocks of financial institutions, and construct a network model based on a threshold approach that captures extremal dependence. Our analysis reveals different dependence structures in the Chinese and U.S. financial systems. By applying the maximum independent set (MIS) from graph theory, we identify a subset of institutions with minimal extremal dependence, facilitating the construction of diversified portfolios resilient to risk contagion. We also compare the performance of our proposed portfolios with that of the market portfolios in the two economies.

Complexity vs Empirical Score

  • Math Complexity: 8.5/10
  • 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 advanced mathematics, including multivariate regular variation, spectral measures, and graph theory (maximum independent set), which is dense and theoretically rigorous. It also demonstrates empirical rigor by applying the methodology to real-world financial data (Chinese and U.S. stocks), constructing networks, running backtests, and comparing portfolio performance against market benchmarks.

Research Flowchart

  flowchart TD
  A["Research Goal<br>Mitigate systemic risk contagion among<br>financial institutions during extreme events"]
  B["Data Inputs<br>Historical stock returns<br>for key institutions"]
  C["Methodology Steps<br>1. Extreme Value Theory<br>2. Extremal Dependence Network<br>3. Maximum Independent Set"]
  D["Computational Processes<br>Construct threshold network<br>Calculate extremal correlations<br>Apply MIS algorithm"]
  E["Key Findings & Outcomes<br>Comparative analysis: CN vs US systems<br>Identification of non-overlapping risk clusters<br>Construction of resilient portfolios"]
  
  A --> B
  B --> C
  C --> D
  D --> E

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 robust portfolio strategy that uses extremal dependence networks and the maximum independent set to mitigate systemic risk during extreme events.

Source sentence
In this paper, we propose a robust portfolio strategy to mitigate systemic risks during extreme events.

Assumes 6

  • Holding period is one day for risk measurement. model
    Source sentence
    We assume the holding period ∆t is one day and calculate the ES of each stock at a 95% confidence level.
  • Portfolio weights are bounded between 0 and 0.1 and sum to 1. execution
    Source sentence
    Constraints are imposed to ensure that the sum of weights equals 1, with each weight coefficient ranging from 0 to 0.1.
  • Expected return of the Chinese portfolio must be at least the three‑month deposit rate of 1.15% set by the Chinese Central Bank. market
    Source sentence
    For the Chinese market, given that the three-month deposit rate in 2024 set by the Chinese Central Bank is 1.15%, we require the expected return of the portfolio to be at least 1.15%.
  • Extremal dependence measure (EDM) accurately captures tail dependence between institution stocks. statistical
    Source sentence
    as our main tool to construct the network structure between stock returns.
Show 2 more
  • The EDM thresholding yields a network where edges represent strong extremal dependence and absence of edges implies little systemic connectivity. model
    Source sentence
    Using the EDM, we construct a network where edges indicate strong extremal dependence between institutions, while the absence of an edge implies little systemic connectivity.
  • Expected shortfall (ES) at 95% confidence is a coherent risk measure suitable for minimizing extremal loss risk. statistical
    Source sentence
    In this section, we propose a portfolio strategy to minimize the risk of extremal loss, where one common systemic risk measure, expected shortfall (ES) is used.

Data

Backtest sample: 2024-01 – 2024-03 · Chinese A‑shares and US S&P 500 stocks · equities · daily

  • Chinese A‑shares stock data · Yahoo Finance via R package quantmod market-prices public
    2023-01-01 to 2023-12-31 · 48 banks and insurance companies
    Source sentence
    ... the chosen time interval spans from January 1, 2023 to December 31, 2023, with a total of 242 trading days.
  • U.S. S&P 500 stock data · Yahoo Finance via R package quantmod market-prices public
    2023-01-01 to 2023-12-31 · 37 banks and insurance companies
    Source sentence
    ... 37 stocks from U.S. S&P 500 in 2023, all of which are banks and insurance companies.

Recognised sources: Yahoo Finance

Costs

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

Finds 5

  • The MIS‑based portfolios achieve lower expected shortfall than market indices in both Chinese and U.S. markets. performance
    Source sentence
    compare the performance of the constructed portfolios in both Chinese and U.S. cases, demonstrating the effectiveness of the proposed strategy.
  • Increasing the EDM threshold creates more isolated vertices and reduces network density, especially in the Chinese market. statistical
    Source sentence
    According to Table 1, in the Chinese market, as the threshold increases, the number of isolated vertices in the network rises while the number of edges decreases, leading to lower average degrees.
  • U.S. financial institutions exhibit higher graph density and clustering, suggesting greater vulnerability to rapid systemic risk spread. statistical
    Source sentence
    The U.S. networks consistently exhibit smaller network diameters, higher graph densities, and higher average clustering coefficients. These characteristics suggest that U.S. financial institutions may be more vulnerable to systemic risks.
  • Betweenness centrality differs between markets; Chinese top institutions have higher betweenness than U.S. counterparts, indicating a more hierarchical structure. statistical
    Source sentence
    The top 8 central financial institutions in the Chinese market have significantly higher betweenness than those in the U.S. market, highlighting the more layered and hierarchical structure of the Chinese system.
  • Key U.S. institutions such as KeyCorp and USBancorp are identified as potentially risky spots in the extremal dependence network. other
    Source sentence
    Figure 6(b) shows that KeyCorp and USBancorp are two potentially risky spots in the U.S. financial system.

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 51 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 (54)

  1. Value at risk: The new benchmark for managing financial risk (2007)
  2. Debt R ank: Too Central to Fail? Financial Networks, the FED and Systemic Risk (2012) — link
  3. Heavy-Tail Phenomena: Probabilistic and Statistical Modeling (2007)
  4. Extremal dependence measure and extremogram: the regularly varying case (2012)
  5. Coherent measures of risk (1999)
  6. Identifying influential energy stocks based on spillover network (2020)
  7. Dynamic asset trees and Black Monday (2003)
  8. Multidimensional minimal spanning tree: The Dow Jones case (2008)
and 46 more
  1. How likely is contagion in financial networks? (2015)
  2. A network analysis of the Chinese stock market (2009)
  3. Portfolio homogeneity and systemic risk of financial networks (2023)
  4. Regularly varying multivariate time series (2009)
  5. Extremes and Local Dependence in Stationary Sequences (1982)
  6. Stable Non-Gaussian Random Processes: Stochastic Models with Infinite Variance (1994)
  7. Statistics for near independence in multivariate extreme values (1996)
  8. The extremogram: A correlogram for extreme events (2009)
  9. The extremal dependence measure and asymptotic independence (2004)
  10. Identifying influential stock indices from global stock markets: A social network analysis approach (2011)
  11. Extreme contagion in equity markets (2004)
  12. Extreme dependence and risk spillovers across north american equity markets (2019)
  13. An Application of extreme value theory for measuring Financial Risk in BRICS Economies (2023)
  14. Correlation and dependence in risk management: properties and pitfalls (2002)
  15. Random Graphs and Complex Networks (2024)
  16. Clustering of Japanese stock returns by recursive modularity optimization for efficient portfolio diversification (2014)
  17. Finding and evaluating community structure in networks (2004)
  18. Fast algorithm for detecting community structure in networks (2004)
  19. Finding community structure in very large networks (2004)
  20. An efficient heuristic procedure for partitioning graphs (1970)
  21. Detecting communities in large networks (2005)
  22. Partitioning sparse matrices with eigenvectors of graphs (1990)
  23. The Theory of Graphs (2001)
  24. Reducibility Among Combinatorial Problems (2010)
  25. Multi-neighborhood tabu search for the maximum weight clique problem (2012)
  26. On the recursive largest first algorithm for graph colouring (2008)
  27. Performance analysis of (1+1) EA on the maximum independent set problem (2015)
  28. Manajemen Risiko Finansial (2007)
  29. A note on vague convergence of measures (2019)
  30. Living on the multidimensional edge: seeking hidden risks using regular variation (2013)
  31. Regular variation for measures on metric spaces (2006)
  32. Heavy-Tailed Time Series (2020)
  33. Regularly Varying Measures on Metric Spaces: Hidden Regular Variation and Hidden Jumps (2014)
  34. Dynamic correlation network analysis of financial asset returns with network clustering (2017)
  35. Community detection in networks: A user guide (2016)
  36. Community detection in temporal multilayer networks, with an application to correlation networks (2016)
  37. Extreme value dependence in financial markets: Diagnostics, models, and financial implications (2004)
  38. Tail risk and asset prices (2014)
  39. Statistical analysis of financial networks (2005)
  40. Chaos based portfolio selection: A nonlinear dynamics approach (2022)
  41. Expected shortfall: a natural coherent alternative to value at risk (2002)
  42. CoVaR (2016)
  43. Measuring Systemic Risk (2016)
  44. Global Systemically Important Insurers: Update 2019 (2019) — link
  45. Manufacturing tail risk: A perspective on the financial crisis of 2007–2009 (2010)
  46. Asset commonality, debt maturity and systemic risk (2012)

0 of 54 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.