Paper: arXiv 2409.12208
Authors: Qian Hui, Tiandong Wang
Abstract
In financial markets marked by inherent volatility, extreme events can result in substantial investor losses. This paper proposes a portfolio strategy designed to mitigate extremal risks. By applying extreme value theory, we evaluate the extremal dependence between stocks and develop a network model reflecting these dependencies. We use a threshold-based approach to construct this complex network and analyze its structural properties. To improve risk diversification, we utilize the concept of the maximum independent set from graph theory to develop suitable portfolio strategies. Since finding the maximum independent set in a given graph is NP-hard, we further partition the network using either sector-based or community-based approaches. Additionally, we use value at risk and expected shortfall as specific risk measures and compare the performance of the proposed portfolios with that of the market portfolio.
Complexity vs Empirical Score
- Math Complexity: 8.5/10
- Empirical Rigor: 6.0/10 (capped from 6.5: transaction costs not modelled; see penalty flags)
- Quadrant: Holy Grail — high math complexity, high empirical rigor (effective empirical rigor capped from 6.5)
Why this score: The paper employs advanced mathematics including extreme value theory, multivariate regular variation, and graph-theoretic NP-hard problems, placing it in the high math category. It also demonstrates empirical rigor by using real market data, implementing a specific algorithm with risk metrics (VaR, ES), and conducting backtest-like performance comparisons.
Research Flowchart
flowchart TD A["Research Goal: Mitigate Extremal Risks in Portfolio Construction"] --> B["Data: Daily Equity Returns & Market Indices"] B --> C["Key Methodology: Extreme Value Theory & Network Modeling"] C --> D["Computational Process: Threshold-Based Network & Structural Analysis"] D --> E["Computational Process: Heuristic Network Partitioning<br>(Sector/Community) & Max Independent Set Approximation"] E --> F["Key Outcome: Proposed Portfolio Strategies"] F --> G["Comparison: Risk-Adjusted Performance<br>vs. Market Portfolio<br>using VaR & Expected Shortfall"]
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 network‑based portfolio strategy that uses extremal dependence measures and maximum independent set selection to mitigate extremal risks.
Source sentence
This paper proposes a portfolio strategy designed to mitigate extremal risks.
Assumes 5
- Holding period is one day for VaR and ES calculations. market
Source sentence
We assume the holding period ∆t is 1 day and calculate the VaR and ES of each stock at a 95% confidence level.
- Portfolio weights sum to one, each between 0 and 0.3, and overall return must be at least 1.15%. execution
Source sentence
Constraints are imposed to ensure that the sum of weights equals 1, with each weight coefficient ranging from 0 to 0.3. Since the current three-month deposit rate set by the Chinese Central Bank is 1.15%, we restrict the desired overall return rate to be at least 1.15%.
- Extremal dependence is measured by the EDM using a threshold‑based network construction. model
Source sentence
Step 1: Compute the log return ri(t) for stock i, and then calculate the pairwise EDM based on Eq.(7); Step 2: Denote each stock as a vertex, and use a threshold‑based approach to construct networks.
- Maximum independent set is approximated using a greedy heuristic. model
Source sentence
In this paper, we use the greedy algorithm in [20] to find solutions to the MISP.
Show 1 more
- Sector‑based and community‑based partitions are used to split the network. model
Source sentence
We compare two methods of partitioning the network: sector‑based and community‑based approaches.
Data
2023-01 – 2023-12 · Shenzhen CSI 300 stocks · equities · daily
- Shenzhen component of the CSI 300 stocks (2023) market-prices public
Source sentence
We use the R package quantmod to retrieve data from Yahoo Finance on 113 constituent stocks from the Shenzhen component of the CSI 300 in 2023. The trading period spans from January 1, 2023 to December 31, 2023, with a total of 242 trading days.
Costs
Not modeled
Finds 5
- Sector‑based portfolios exhibit more volatile returns and higher risk than community‑based portfolios. performance
Source sentence
portfolio has more volatile returns and risks than the community‑based one, so we conclude that partitioning the dependence network by sectors is suboptimal.
- Higher network thresholds produce more isolated vertices, lower average degree, and increase diameter and path length up to a point. statistical
Source sentence
... network diameter and path length peak at θ = 0.15, but then decline as the number of isolated vertices significantly disrupts the network structure.
- After extracting sub‑networks, many isolated vertices appear, indicating strong extremal dependence across sectors. statistical
Source sentence
Figure 4 reveals that after extracting sub‑networks, a large number of isolated vertices appear, whereas the original network only contains 13 isolated vertices. This phenomenon suggests strong extremal dependence among different sectors.
- The greedy heuristic can find feasible independent sets for each sector or community to construct portfolios. theoretical
Source sentence
This algorithm ensures that the solution set we find is an independent one by gradually expanding the vertex set until all possibilities are exhausted to obtain a feasible solution.
- The proposed network‑based portfolio strategy, using VaR and ES minimization, reduces exposure to extremal losses compared with the market portfolio. performance
Source sentence
In this section, we propose a portfolio strategy to minimize the risk of extremal loss, where two common risk measures, value‑at‑risk (VaR) and expected shortfall (ES) are used.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 28 related papers). Coverage is partial, so this is neither replication nor refutation.
1 paper in the archive : A Survey of Financial AI: Architectures, Advances and Open …
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 (47)
- Heavy-Tail Phenomena: Probabilistic and Statistical Modeling (2007)
- Extremal dependence measure and extremogram: the regularly varying case (2012)
- Coherent measures of risk (1999)
- Identifying influential energy stocks based on spillover network (2020)
- Dynamic asset trees and Black Monday (2003)
- Multidimensional minimal spanning tree: The Dow Jones case (2008)
- How likely is contagion in financial networks? (2015)
- A network analysis of the Chinese stock market (2009)
and 39 more
- Portfolio homogeneity and systemic risk of financial networks (2023)
- Regularly varying multivariate time series (2009)
- Extremes and Local Dependence in Stationary Sequences (1982)
- Stable Non-Gaussian Random Processes: Stochastic Models with Infinite Variance (1994)
- Statistics for near independence in multivariate extreme values (1996)
- The extremogram: A correlogram for extreme events (2009)
- The extremal dependence measure and asymptotic independence (2004)
- Identifying influential stock indices from global stock markets: A social network analysis approach (2011)
- Extreme contagion in equity markets (2004)
- Extreme dependence and risk spillovers across north american equity markets (2019)
- An Application of extreme value theory for measuring Financial Risk in BRICS Economies (2023)
- Correlation and dependence in risk management: properties and pitfalls (2002)
- Random Graphs and Complex Networks (2024)
- Clustering of Japanese stock returns by recursive modularity optimization for efficient portfolio diversification (2014)
- Finding and evaluating community structure in networks (2004)
- Fast algorithm for detecting community structure in networks (2004)
- Finding community structure in very large networks (2004)
- An efficient heuristic procedure for partitioning graphs (1970)
- Detecting communities in large networks (2005)
- Partitioning sparse matrices with eigenvectors of graphs (1990)
- The Theory of Graphs (2001)
- Reducibility Among Combinatorial Problems (2010)
- Multi-neighborhood tabu search for the maximum weight clique problem (2012)
- On the recursive largest first algorithm for graph colouring (2008)
- Performance analysis of (1+1) EA on the maximum independent set problem (2015)
- Manajemen Risiko Finansial (2007)
- A note on vague convergence of measures (2019)
- Living on the multidimensional edge: seeking hidden risks using regular variation (2013)
- Regular variation for measures on metric spaces (2006)
- Heavy-Tailed Time Series (2020)
- Regularly Varying Measures on Metric Spaces: Hidden Regular Variation and Hidden Jumps (2014)
- Dynamic correlation network analysis of financial asset returns with network clustering (2017)
- Community detection in networks: A user guide (2016)
- Community detection in temporal multilayer networks, with an application to correlation networks (2016)
- Extreme value dependence in financial markets: Diagnostics, models, and financial implications (2004)
- Tail risk and asset prices (2014)
- Statistical analysis of financial networks (2005)
- Chaos based portfolio selection: A nonlinear dynamics approach (2022)
- Expected shortfall: a natural coherent alternative to value at risk (2002)
0 of 47 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.