Replication Data and Empirical Results for Institutional Portfolio Optimization under Cardinality and Execution Frictions

This dataset contains the empirical input data, walk-forward out-of-sample portfolio allocations, and statistical evaluation results for the manuscript titled: "Institutional Portfolio Optimization under Cardinality and Execution Frictions: A Walk-Forward Evaluation of Adaptive Diversity Preservation". Contents: sp500_daily.csv: Point-in-time daily simple returns and trading volume panel for S&P 500 constituent equities spanning 2012-01-04 to 2025-01-30. ken_french_carhart_daily.csv: Daily Fama-French/Carhart factors (Market, SMB, HML, Momentum, and Risk-Free Rate) used for econometric performance attribution. results_replication.zip: Complete set of 15 CSV files and 4 high-resolution figures documenting the 7-year expanding walk-forward horizon (2018–2024), including baseline comparisons (SOS, GA, PSO, DE, Ledoit-Wolf, 1/N), component ablation studies, CSCV/PBO overfitting matrices, Almgren-Chriss capacity curves up to $5B AUM, and volatility regime partitions. Code Repository: Replication code to regenerate all models and metrics is available on GitHub at: https://github.com/atishayj2202/Intelligent-Computing-Portfolio-Optimization.