Paper: arXiv 2402.01951
Authors: Stelios Arvanitis, Olivier Scaillet, Nikolas Topaloglou
Abstract
We develop and implement methods for determining whether relaxing sparsity constraints on portfolios improves the investment opportunity set for risk-averse investors. We formulate a new estimation procedure for sparse second-order stochastic spanning based on a greedy algorithm and Linear Programming. We show the optimal recovery of the sparse solution asymptotically whether spanning holds or not. From large equity datasets, we estimate the expected utility loss due to possible under-diversification, and find that there is no benefit from expanding a sparse opportunity set beyond 45 assets. The optimal sparse portfolio invests in 10 industry sectors and cuts tail risk when compared to a sparse mean-variance portfolio. On a rolling-window basis, the number of assets shrinks to 25 assets in crisis periods, while standard factor models cannot explain the performance of the sparse portfolios.
Complexity vs Empirical Score
- Math Complexity: 8.5/10
- Empirical Rigor: 7.0/10
- Quadrant: Holy Grail — high math complexity, high empirical rigor
Why this score: The paper employs advanced mathematical constructs including second-order stochastic dominance, submodular optimization, and asymptotic recovery proofs (high math complexity), while also presenting large-scale empirical backtests on equity datasets with rolling-window analysis and tail risk metrics (high empirical rigor).
Research Flowchart
flowchart TD A["Research Goal: Determine if relaxing<br>sparsity constraints improves<br>investment opportunity sets"] --> B["Methodology: Greedy Algorithm &<br>Linear Programming for Sparse<br>Second-Order Stochastic Spanning"] B --> C["Large Equity Datasets<br>(Rolling Windows)"] C --> D["Computational Process:<br>Estimate Optimal Sparse Portfolios<br>from 1 to 100+ Assets"] D --> E["Key Findings 1:<br>45 assets optimal; 10 sectors<br>captured; no benefit beyond 45"] D --> F["Key Findings 2:<br>25 assets in crises; cuts tail risk<br>vs. Mean-Variance; unexplained<br>by standard factor models"]
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 introduces a new estimation procedure for sparse second‑order stochastic dominance spanning using a greedy algorithm and linear programming.
Source sentence
We formulate a new estimation procedure for sparse second-order stochastic spanning based on a greedy algorithm and Linear Programming.
Assumes 1
- The returns process (X_t) is strictly stationary, absolutely regular with mixing coefficients decaying geometrically, and has finite (2+ε) moments. statistical
Source sentence
Assumption 2. (X∞ t )t∈Z is strictly stationary and absolutely regular with mixing coefficients (βm)m∈N that satisfy βm ∼bm for some b ∈(0, 1), as m →∞, and max0<i≤+∞E h X(i)
Data
- S&P 500 constituents market-prices public
Source sentence
period from January 1981 to December 2020, namely a total of 480 monthly return observations.
Costs
Modeled
A proportional transaction cost of 35 basis points is applied to portfolio turnover.
Source sentence
We use a transaction cost of 35 bps, which is typical in the literature.
Finds 5
- Expanding the sparse opportunity set beyond 45 assets yields no additional benefit for risk‑averse investors. theoretical
Source sentence
find that there is no benefit from expanding a sparse opportunity set beyond 45 assets.
- The optimal sparse portfolio uses 10 industry sectors and reduces tail risk relative to a sparse mean‑variance portfolio. performance
Source sentence
The optimal sparse portfolio invests in 10 industry sectors and cuts tail risk when compared to a sparse mean-variance portfolio.
- During crisis periods the optimal sparse portfolio shrinks to about 25 assets. performance
Source sentence
On a rolling-window basis, the number of assets shrinks to 25 assets in crisis periods.
- SS‑SSD portfolios have higher average returns, Sharpe ratios, downside Sharpe ratios and lower VaR/ES than MAXSER portfolios. performance
Source sentence
We observe that the Average, the Sharpe Ratios and the Downside Sharpe Ratios of the SS-SSD optimal portfolios are higher than those of the MAXSER optimal portfolios.
- Statistical test rejects non‑dominance of MAXSER over SS‑SSD (p‑value 4.4%). statistical
Source sentence
The test statistic ξT is -0.0012, and the p-value is estimated at 4.4%. We thus reject the null hypothesis of non-dominance of portfolio SS-SSD over MAXSER.
Checked by others
No other paper in the archive has been found to support or contradict these findings (compared with 44 related papers). Coverage is partial, so this is neither replication nor refutation.
3 papers in the archive : Sizing the bets in a focused portfolio
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.