Practical Problems with Mean-Variance Optimization and the Black-Litterman Solution
While Markowitz's mean-variance optimization is theoretically elegant, it suffers from a serious practical problem: it produces extreme asset allocations. Changing the estimated expected returns by even a small amount can cause the optimal portfolio composition to shift dramatically, sometimes resulting in unrealistic outcomes such as 100% concentration in a single asset class. The Black-Litterman model (BL model), developed in 1990 by Fischer Black and Robert Litterman at Goldman Sachs, solved this problem by introducing market equilibrium returns as a starting point.
The core idea of the BL model is to use market equilibrium returns (implied returns) derived from the CAPM as a prior distribution, and then integrate the investor's subjective views through Bayesian estimation. Market equilibrium returns are the expected returns for each asset reverse-engineered under the assumption that the current market-cap-weighted portfolio is the optimal portfolio. Using this starting point, the model recommends the market portfolio when the investor has no views, and deviations occur only when views are expressed - producing intuitive results.
Setting Views and Expressing Confidence Levels
The most important step in practicing the BL model is expressing investor views quantitatively. There are two types of views: "absolute views" and "relative views." An absolute view predicts the return of a specific asset class, such as "Japanese equities will rise 8% over the next year." A relative view predicts the relative return difference between two asset classes, such as "US equities will outperform European equities by 3%." Relative views tend to be more accurate than absolute level predictions and are widely used in practice.Books on Bayesian estimation and investment forecasting also explain that properly setting the confidence level of views is what determines the quality of the model's output.
Each view is assigned a confidence level. Views with higher confidence have a greater impact on the final portfolio. Views with low confidence result in only minor adjustments to the market equilibrium returns. This mechanism allows investors to distinguish between "views held with strong conviction" and "views that are merely hunches" and reflect them accordingly in the model. Setting confidence levels takes into account the track record of past prediction accuracy and the quality of data underlying the views.
How Individual Investors Can Apply the BL Model's Thinking
While the mathematical implementation of the BL model is complex, the thinking behind it is fully applicable for individual investors. First, adopt the market portfolio (such as a global equity index) as your starting point. Then, adjust the portfolio only when you have views in which you are confident. For example, if you have high conviction that "emerging market growth rates will exceed those of developed markets," set the emerging market equity allocation slightly above market weight.
The key discipline is to maintain market weights for asset classes about which you have no views.Books on market portfolios and index investing emphasize that the approach of eliminating unfounded biases and deviating from the market only in areas of conviction is a rational method that reflects active views while suppressing the risk of excessive concentration. The philosophy of the BL model is about bringing a "balance of humility and conviction" to investing.
Next Actions for Applying BL Model Thinking to Your Investments
To put the BL model's thinking into practice, start by checking how much your current portfolio deviates from the market portfolio. Compare the regional composition of a global equity index (approximately 60% US, 15% Europe, 5% Japan, 10% emerging markets, etc.) with your own portfolio, and determine whether deviations are intentional or unconscious biases. If you find intentional deviations, document the underlying views and self-assess your confidence on a 1-5 scale.
If you find large deviations based on low-confidence views, consider adjusting your portfolio closer to market weights. Use our compound interest calculator to run comparison simulations of long-term returns between a portfolio close to market weights and your current portfolio to quantitatively understand the impact of deviations. Establish a cycle of reviewing your views and confidence levels every six months and reflecting any changes in your portfolio.