量化研究系列报告之二十五：高弹性Alpha的量化掘金：从盲区识别到策略构建

Group 1 - The report highlights the limitations of traditional multi-factor models, which face inherent path dependency and structural mismatches, leading to a dilution of returns and an inability to capture high elasticity styles [2][25][26] - The report proposes a dual-driven solution based on XGBoost non-linear prediction and high elasticity alpha extraction, achieving an annualized excess return of 20.0% across ten market segments with an information ratio of 3.78 [3][4] - The integration of high elasticity strategies significantly enhances the performance of traditional index-enhanced models, with annualized excess returns improving by 2.1% to 4.7% compared to single strategies [4][12][19] Group 2 - The report discusses the challenges faced by traditional multi-factor models, particularly their reliance on historical data and the inability to adapt to changing market structures, which can lead to systematic failures during specific market conditions [21][22][25] - It emphasizes the non-normal distribution of returns in the market, where excess returns are often concentrated in a few stocks, contradicting the diversification philosophy of traditional models [26][28][29] - The analysis reveals that the performance of quantitative strategies is closely tied to specific style factors, indicating a path dependency that can hinder adaptability in dynamic market environments [32][34][37]