Robust Optimization of Strategic and Tactical Asset Allocation for Multi-Asset Portfolios
针对包含流动与非流动资产的现代多资产组合，提出统一鲁棒优化框架ROSAA，通过多资产可交易因子和分层聚类组Lasso方法，实现稳健协方差估计与战略战术配置的衔接。
Traditional asset allocation frameworks face significant challenges when applied to modern multi-asset portfolios that include liquid public assets alongside illiquid private assets and hedge funds. Mixed liquidity profiles, smoothed or appraisal-based valuations, and high dimensionality make robust covariance estimation and coherent portfolio construction difficult using standard approaches. We develop a unified robust optimization of strategic and active asset allocation (ROSAA) framework linking strategic asset allocation (SAA) and tactical asset allocation (TAA) for heterogeneous asset universes. First, we introduce multi-asset tradable factors (MATF) constructed from liquid futures and quantitative strategy trackers to model cross-asset risk premia, including a tradable private equity risk premia factor for consistent treatment of private assets. Second, we propose hierarchical clustering group Lasso (HCGL) for stable and interpretable covariance estimation, combining correlation-based hierarchical clustering with multivariate group Lasso regularization. HCGL incorporates economically motivated sign and zero-exposure constraints, ensuring meaningful factor loadings across asset classes and preventing spurious hedging. Within ROSAA, the SAA is determined using risk-budgeting optimization that avoids reliance on capital market assumption forecasts. The TAA maximizes allocations to tactical alpha characteristics, including volatility-normalized momentum and low-beta alphas for traditional long-only assets, and factor-adjusted manager alphas for alternative investments, subject to tracking error and turnover controls.