Core Insights - Asset allocation is not just about risk diversification; predicting returns is equally important. The report emphasizes that the significance of asset allocation lies in systematically designing strategies that allow investors to leverage their predictive abilities, enabling those with a 60% win rate to outperform those with a 40% win rate [5][8] - The allocation approach can be categorized into passive and active types. Passive strategies focus on risk diversification, while active strategies, such as Mean-Variance Optimization (MVO) and Black-Litterman (B-L), aim to enhance returns while considering risk [5][8] - Active asset allocation's core is return prediction and risk penalty, where risk is defined as "uncertainty." For professional investors, asset volatility is not risk if they can predict it; thus, true risk stems from inadequate predictive ability [5][11] MVO Model Application - The MVO model is suitable for active asset allocation, with its derivatives like the B-L model incorporating subjective views on returns. However, the model is sensitive to input variables, which can lead to concentrated positions in a few asset classes [9][10] - The MVO model requires high accuracy in return predictions, which is both a limitation and a value. It allows investors with a predictive edge to construct effective asset allocation strategies [10][29] Active Asset Allocation Scenarios - Active asset allocation can be applied in two scenarios: directly obtaining allocation schemes or enhancing passive allocations with active strategies. The latter involves using passive models to establish a base and then applying active strategies to enhance returns on assets where predictive capabilities exist [36][43] ETF-Based Active Asset Allocation - The report discusses an ETF-based active asset allocation strategy, replacing traditional indices with investable ETFs. This includes using an industry rotation strategy for equities and bond ETFs to replicate mixed bond fund indices [49][51] - The active ETF allocation strategy can be categorized into two types: one directly based on MVO and the other combining a passive strategy with MVO enhancements [59][60]