LOADING
Portfolio construction is the process of translating investment signals into actual portfolio positions. Even the most brilliant stock selection model will underperform if the portfolio construction step introduces unintended risks, excessive concentration, or unnecessary turnover. The decisions made at this stage, including weighting scheme, rebalancing frequency, and constraint specification, have a profound impact on realized returns and risk characteristics.
The simplest weighting approach is equal weighting, which assigns the same dollar amount to each position. Equal weighting is straightforward, avoids concentration in a few large positions, and implicitly tilts toward smaller-cap stocks. However, it requires frequent rebalancing as position sizes drift, and it gives the same weight to your strongest conviction ideas as to your weakest. Market-cap weighting, used by most index funds, weights positions by their market capitalization. It minimizes turnover but concentrates the portfolio in the largest companies.
Optimization-based approaches use mathematical programming to determine portfolio weights that maximize expected return for a given level of risk (or minimize risk for a given expected return). Mean-variance optimization, introduced by Markowitz, is the foundational approach. In practice, however, MVO is highly sensitive to estimation errors in expected returns and covariances, often producing extreme, unintuitive portfolios. Robust optimization techniques, such as shrinkage estimators for the covariance matrix and Black-Litterman for expected returns, help produce more stable and realistic portfolios.
Constraints are essential for practical portfolio construction. Typical constraints include maximum position size (e.g., no single stock exceeds 5% of the portfolio), sector exposure limits, minimum number of holdings, and turnover limits. These constraints prevent the optimizer from producing degenerate solutions and keep the portfolio aligned with its stated investment philosophy. Risk factor exposure constraints (such as limiting beta deviation from the benchmark) ensure the portfolio does not take unintended macro bets.
Rebalancing frequency involves a tradeoff between signal freshness and transaction costs. More frequent rebalancing (daily or weekly) keeps the portfolio closer to target weights but incurs higher trading costs. Less frequent rebalancing (quarterly or annually) reduces costs but allows positions to drift significantly. The optimal frequency depends on the alpha decay rate of the underlying signals: fast-decaying signals (like short-term momentum) require more frequent rebalancing than slow-moving signals (like annual value metrics).