Finance Theory — 14.6: Risk Parameter Estimation and Stationarity

Описание: How much historical data should you use to estimate portfolio risk parameters — and when should you override the math with judgment? This video walks through the covariance estimation formula with a concrete numerical example, explains the fundamental tradeoff between statistical precision and economic relevance, and builds a practical decision framework for choosing estimation windows.

Key concepts covered:
• Sample covariance formula: computing cross-products of return deviations and averaging across periods
• Numerical example: two-stock, four-period covariance calculation step by step
• Non-stationarity: why the statistical properties of financial returns change over time
• The estimation window tradeoff: more data reduces noise, but older data introduces systematic bias
• Annualizing volatility: converting monthly variance to annual variance using the square root of 12 rule
• Portfolio theory vs. stock picking: two fundamentally different investment philosophies and their assumptions
• Why crisis data (1987, 1998, 2000, 2008, 2020) should not be excluded from long-term estimates
• Rolling correlations: how S&P 500 vs. NASDAQ correlation shifts dramatically during crises
• Practitioner decision framework: structural breaks, regime dependence, and investment horizon
• The Venn diagram of effective risk estimation: mathematical framework, economic understanding, and contextual judgment

ORIGINAL SOURCE
This video is based on content from the following source:
   • Ses 14: Portfolio Theory II