| 摘要: | 本研究針對台股結構與散戶投資環境，提出一套中長線量化選股策略，透過財報因子敏感度探勘與績效分層分析，提升策略的在地適應性與穩健性。
本策略整合財務穩健度（F）、成長動能（C）、估值條件（V）三構面，建構具邏輯層次的多因子選股架構，並以 ROE、自由現金流、總資產周轉率與稅後淨利率等指標為核心篩選依據。
為驗證策略成效，設計具資料對齊、篩選與回測功能之量化系統，使用 2006–2024 年台股月資料進行實證分析。結果顯示，C3（ROE 高點）為最具主導性的序列變動指標，F1F2 的財務條件展現良好穩健性，V 構面（本益比）則發揮調節效果，並經 ANOVA 與非參數檢定驗證具統計顯著性。
研究建議策略可採「C 主軸、F 篩選、V 調節」邏輯，並貢獻於構面導向的因子設計、序列型動能指標建構，以及結合視覺化與統計檢定的績效說明機制。
限制包括因子範圍、市場樣本與區域性侷限。未來可導入機器學習加權法、擴展跨市場驗證，並結合情緒訊號與產業異質性分析，以提升策略泛用性與即時反應能力。;This study proposes a mid-to-long-term quantitative stock selection strategy tailored to the structure of Taiwan’s equity market and retail investor behavior. By exploring financial factor sensitivity and performance stratification, the strategy enhances local adaptability and robustness.
The framework integrates three dimensions—Financial Stability (F), Corporate Growth Momentum (C), and Valuation Conditions (V)—to construct a multi-factor selection logic. Core indicators such as ROE, free cash flow, total asset turnover, and net profit margin are used as screening criteria.
To validate performance, a backtesting system with data alignment, filtering, and statistical evaluation functions was developed, using Taiwan’s monthly stock and financial data from 2006 to 2024. Empirical results show that C3 (ROE peak) is the most influential sequence-based indicator, while the F1F2 combination demonstrates solid robustness. The valuation factor (V) plays a moderating role, and strategy differences are statistically significant under ANOVA and nonparametric tests.
The study recommends a "C-driven, F-filtered, V-adjusted" logic, contributing to dimension-based factor design, sequence-based momentum indicators, and an explanatory mechanism integrating visualization and statistical validation.
Limitations include factor scope, data coverage, and regional specificity. Future research may incorporate machine learning-based factor weighting, cross-market validation, sentiment signals, and industry heterogeneity analysis to enhance generalizability and responsiveness. |
