Research
This archive covers portfolio construction, backtesting, implementation, and other quantitative research that sits alongside the dedicated signals library.
Point-in-time data handling before performance interpretation.
Validation that survives turnover, cost, and capacity assumptions.
Implementation detail treated as part of the research, not cleanup.
Clear documentation of model scope, limits, and decision rules.
Search research
Showing 5 of 5 articles.
A raw minimum variance portfolio often looks safer because it is simply running less risk. A better comparison is to scale it up to the volatility of equal weight, then make the borrowing cost explicit.
Minimum variance weighting minimizes total portfolio volatility by using the full covariance matrix, making diversification a system-level optimization problem rather than a one-asset-at-a-time rule.
A step-by-step derivation of the portfolio variance formula, showing why portfolio risk is written as wᵀΣw and why correlations matter as much as individual asset volatility.
More from the archive
Portfolio Construction · March 29, 2026
Backtesting · March 18, 2026
Page 1 of 1
