Replication Data for Selection Confidence Sets for Equally Weighted Portfolios

Folder 1: MC_SIMULATIONS Contains R scripts and output files (.RData) for all Monte Carlo experiments reported in the paper and Supplementary Material. Organized in four subfolders: MODEL_1 (graph-based dependence, v in {-1, -0.2}) and MODEL_2 (exchangeable correlation, rho in {0.75, 0.25}) for the asymptotic Wald-test SCS; BOOTSTRAP for the nonparametric bootstrap-based SCS; LEDOIT_&_WOLF for SCSs constructed via the Ledoit & Wolf bootstrap test for Sharpe ratio differences. Each configuration covers three loss functions: Sharpe ratio (SR), mean-variance (MEANVAR), and expected shortfall (ES). Folder 2: REAL_DATA Contains the datasets and the R scripts used for the empirical applications (L1-Cryptocurrency and French 17 Industry Portfolios) presented in the paper.

文件夹1：蒙特卡洛模拟（MC_SIMULATIONS）
包含本文及补充材料中所报告的全部蒙特卡洛实验的R脚本与.RData格式输出文件。该文件夹下设四个子文件夹：
1. 模型1（MODEL_1）子文件夹：针对基于图的相关性结构，v取值为{-1, -0.2}，用于渐近沃尔德检验SCS方法；
2. 模型2（MODEL_2）子文件夹：针对可交换相关性结构，rho取值为{0.75, 0.25}，用于渐近沃尔德检验SCS方法；
3. 自助法子文件夹（BOOTSTRAP）：存放基于非参数Bootstrap的SCS方法相关文件；
4. 勒多伊特-沃尔夫（Ledoit & Wolf）子文件夹：存放通过针对夏普比率差异的勒多伊特-沃尔夫Bootstrap检验构建的SCS方法相关文件。
每种配置均涵盖三类损失函数：夏普比率（Sharpe Ratio, SR）、均值-方差（MEANVAR）与期望短缺（Expected Shortfall, ES）。

文件夹2：实证数据（REAL_DATA）
包含本文展示的实证分析应用所用的数据集与R脚本，涉及L1加密货币数据集与法国17行业投资组合两组实证场景。