Paper: arXiv 2611.21897
Authors: Farzana Afroz, Abootaleb Shirvani, Nicholas Appiah, Svetlozar T. Rachev, Frank J. Fabozzi
Abstract
This study examines thirty equity–bond and multi-asset exchange-traded funds (ETFs) listed in the United States, Canada, and Australia using daily U.S.-dollar returns over 2020–2024. A rolling-window framework compares minimum-variance and Conditional Value-at-Risk (CVaR) portfolio optimization under long-only and long–short constraints against an equally weighted benchmark. Performance is assessed using risk-adjusted measures, downside risk, extreme-tail estimation, and conditional volatility models. Optimized portfolios exhibit lower empirical VaR and CVaR than the benchmark, with long–short strategies generally providing greater downside-risk reduction. However, the equally weighted benchmark achieves the highest reported Sharpe, Sortino, and Rachev ratios. GARCH and FIGARCH comparisons favor GARCH based on information criteria. A historically estimated Normal Inverse Gaussian model further generates European option prices and model-implied volatility surfaces under a martingale-corrected pricing specification. The findings highlight trade-offs between risk reduction, risk-adjusted performance, and practical implementation costs.
Complexity vs Empirical Score
- Math Complexity: 7.5/10
- Empirical Rigor: 6.0/10 (capped from 8.0: transaction costs only partially modelled; see penalty flags)
- Quadrant: Holy Grail — high math complexity, high empirical rigor (effective empirical rigor capped from 8.0)
Why this score: The paper demonstrates strong empirical rigor through its extensive backtesting framework, diverse dataset, and robust statistical methodologies. It also incorporates advanced mathematical concepts like Lèvy processes and NIG distributions for option pricing, balancing theoretical depth with practical application. The novelty lies in combining these elements to re-evaluate diversification under recent market conditions.
Research Flowchart
flowchart TD
A["Research Goal: Revisit Equity-Bond Diversification under Tail Risk and Structural Market Change"] --> B{"Key Methodology Steps"};
B --> C["Data/Inputs: 30 Global Equity-Bond ETFs (2020-2024 Daily USD Returns)"];
C --> D{"Computational Processes"};
D --> E["Portfolio Optimization: Minimum-Variance & CVaR (Long-Only/Long-Short)"];
E --> F["Performance Assessment: Risk-Adjusted Measures, Downside Risk, Extreme-Tail, Conditional Volatility (GARCH/FIGARCH)"];
F --> G["Option Pricing: Normal Inverse Gaussian Model & Model-Implied Volatility"];
G --> H["Key Findings/Outcomes"];
H --> I["Optimized Portfolios (Long-Short) Lower Empirical VaR/CVaR; Equally Weighted Benchmark Highest Sharpe/Sortino/Rachev Ratios; GARCH Favored over FIGARCH; Trade-offs between Risk Reduction, Risk-Adjusted Performance, and Implementation Costs."];
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 develops a unified empirical framework that combines rolling‑window portfolio optimization (minimum‑variance and CVaR, long‑only and long‑short), downside‑risk and extreme‑tail evaluation, conditional volatility modeling, and NIG Lévy‑based option‑pricing to study global equity‑bond and mu
Source sentence
Our contribution is fourfold. Fourth, we develop a model-based option-pricing analysis under the NIG L´evy framework to complement the historical portfolio and risk analysis.
Assumes 6
- Daily returns are computed from closing prices expressed in U.S. dollars, assuming that differences in market closing times do not materially affect the analysis. market
Source sentence
Daily ETF closing prices for funds listed in the United States, Canada, and Australia were expressed in U.S. dollars before calculating returns. Differences in market closing times were not explicitly adjusted for, as the analysis was con-ducted using daily observations rather than intraday synchronized prices.
- The AR(1)–GARCH(1,1) model with Gaussian innovations adequately captures serial dependence and conditional volatility for the filtered extreme‑tail analysis. model
Source sentence
For each return series, an AR(1)–GARCH(1,1) model with Gaussian innovations is first fitted to account for serial dependence in returns and time-varying volatility.
- The extreme upper tail of loss magnitudes follows a regularly varying Pareto‑type distribution. statistical
Source sentence
In our analysis, we adopt a Pareto-type tail assumption for these positive loss magnitudes. More specifically, we assume that the extreme upper tail of the loss magnitudes obtained from the negative standardized residuals is regularly varying and can be approximated asymptotically by a power-law relationship of the form P(L > x) ∼Cx−α, x →∞, where L denotes the positive magnitude of a negative standardized residual,
- Short‑borrowing fees and financing costs are ignored in the reported headline portfolio performance. execution
Source sentence
Transaction costs are exam-ined separately through a sensitivity analysis, while short-borrowing fees and financing costs are not explicitly modeled.
Show 2 more
- Transaction costs are proportional and applied equally to purchases and sales during rebalancing. execution
Source sentence
We consider proportional transaction costs of 0, 5, 10, and 20 basis points per dollar traded, applied to both purchases and sales during portfolio rebalancing.
- The NIG parameters estimated from historical EWP returns are suitable for martingale‑corrected European option pricing without further market calibration. model
Source sentence
The estimated NIG model is incorporated into a martingale-corrected pricing specification to obtain European option prices across different strikes and maturities and construct the corresponding implied-volatility surface. Since the NIG parameters are estimated from historical EWP returns rather than calibrated to observed option prices, the resulting option prices and implied volatilities are model-implied rather
Data
2020 – 2024 · global equity‑bond ETFs (US, Canada, Australia) · multi-asset · daily
- Equity‑bond and multi‑asset ETFs daily adjusted closing prices market-prices public
Source sentence
2.1 Data Description The empirical analysis is based on a sample of thirty equity–bond and multi-asset exchange-traded funds listed in three developed financial markets: the United States, Canada, and Australia. Daily adjusted closing prices are obtained from Yahoo Finance over the period January 2, 2020 through December 31, 2024.
- Canadian‑US exchange‑rate series (DEXCAUS) macro public
Source sentence
Canadian-listed ETF prices were converted using the DEXCAUS daily exchange-rate series (Canadian dollars per U.S. dollar), while Australian-listed ETF prices were converted using the DEXUSAL series (U.S. dollars per Australian dollar), both obtained from the Federal Reserve Economic Data (FRED) database maintained by the Federal Reserve Bank of St. Louis (Board of Governors of the Federal Reserve System (US), 2026a,
- Australian‑US exchange‑rate series (DEXUSAL) macro public
Source sentence
Canadian-listed ETF prices were converted using the DEXCAUS daily exchange-rate series (Canadian dollars per U.S. dollar), while Australian-listed ETF prices were converted using the DEXUSAL series (U.S. dollars per Australian dollar), both obtained from the Federal Reserve Economic Data (FRED) database maintained by the Federal Reserve Bank of St. Louis (Board of Governors of the Federal Reserve System (US), 2026a,
- Dow Jones Industrial Average (DJIA) index values benchmark public
Source sentence
In addition to the ETF universe, the Dow Jones Industrial Average (DJIA) is included as a benchmark representing broad U.S. equity market performance. The DJIA serves as a useful reference point for comparing the behavior of balanced portfolios with concentrated large-cap equity exposure during periods of market turbulence and recovery.
- 3‑month U.S. Treasury yield macro public
Source sentence
For the risk-adjusted measures requiring a risk-free rate, the 3-month U.S. Treasury yield is used as the risk-free-rate proxy and converted to an effective daily rate to match the frequency of the portfolio returns.
- 10‑year U.S. Treasury yield macro public
Source sentence
Figures 3 and 4 present the full-sample (in-sample) efficient frontiers obtained under the mean–variance and CVaR optimization frameworks using the 3-month U.S. Treasury yield as a short-term risk-free-rate proxy and the 10-year U.S. Treasury yield as an alternative reference yield.
Costs
Partial
Proportional transaction costs of 0, 5, 10, and 20 basis points per dollar traded are applied to both purchases and sales in a sensitivity analysis.
Source sentence
We consider proportional transaction costs of 0, 5, 10, and 20 basis points per dollar traded, applied to both purchases and sales during portfolio rebalancing.
Finds 5
Sharpe -0.04 (benchmark -0.02) · max drawdown 16.5%
- All six optimized portfolios exhibit lower historical VaR and CVaR than the equally weighted benchmark at both 95% and 99% confidence levels. performance figure from elsewhere in the paper
Source sentence
Taken together, the results show that all six optimized portfolios exhibit lower historical VaR and CVaR estimates than EWP at both confidence levels.
- Long‑short specifications have the smallest tail‑risk estimates, with LS‑MVP achieving the lowest CVaR at both confidence levels. performance
Source sentence
Thus, the long–short portfolio specifications are associated with lower measured downside-tail risk in this sample. The long–short strategies record the smallest tail-risk estimates, with LS-MVP achieving the lowest CVaR at both confidence levels.
- The equally weighted benchmark records the highest Sharpe, Sortino, and Rachev ratios, although all reported Sharpe and Sortino ratios are negative. performance
Source sentence
However, the equally weighted benchmark achieves the highest reported Sharpe, Sortino, and Rachev ratios. However, the EWP achieves the highest reported Sharpe, Sortino, and Rachev ratios, with all reported Sharpe and Sortino ratios remaining negative.
- GARCH models are preferred over FIGARCH based on information‑criteria comparisons. statistical
Source sentence
GARCH and FIGARCH comparisons favor GARCH based on information criteria.
- At a 20‑basis‑point transaction‑cost level, LS‑MVP still shows the lowest daily volatility (0.0025570) compared with the equally weighted portfolio (0.0062243). performance
Source sentence
At 20 basis points, LS-MVP continues to exhibit the lowest daily volatility (0.0025570), compared with 0.0062243 for EWP.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 23 related papers). Coverage is partial, so this is neither replication nor refutation.
Code
No repository by the authors found in the paper.
Extracted 2026-11-18 · same card as JSON · A language model chose these items: check the source sentence before relying on one.