When should I use Black-Litterman instead of Mean-Variance Optimization?

Use Mean-Variance Optimization (MVO) when you have high-confidence return estimates AND you're willing to accept potentially extreme allocations. Use Black-Litterman when you want a more diversified, robust portfolio that incorporates your views without dominating the equilibrium.

MVO in detail:

The MVO sensitivity problem:

If you're optimising across 10 asset classes and you tweak the expected return of one class by 0.5%, the optimal allocation can swing by 30%+. This is mathematically correct but practically useless — no one has 0.5%-precision return estimates.

Black-Litterman in detail:

The Black-Litterman fix:

Instead of starting from scratch, Black-Litterman:

1. Computes equilibrium returns from market cap weights (reverse-MVO trick)
2. Lets you express views about specific assets or spreads
3. Weights views by your confidence in each
4. Blends equilibrium and views into a single return vector
5. Runs MVO on the blended returns

The result: a portfolio that tilts toward your views but doesn't crash on extreme estimates.

When MVO is appropriate:

• High-confidence return estimates (rare in practice)
• Small number of asset classes (5 or fewer)
• Academic / classroom use cases
• Stress-testing extreme allocations

When Black-Litterman is preferred:

• Real-world institutional portfolios
• Many asset classes (10+)
• You have specific views but not on all assets
• You want a robust, diversified output
• You want to avoid corner solutions

Practical comparison for a 60/40 stock/bond starting portfolio:

Suppose equilibrium returns are 7% stocks, 4% bonds. You believe stocks will outperform by 2 percentage points more than the equilibrium suggests.

MVO: Adjust your stock expected return to 9%. With the same risk tolerance, MVO might push you to 90% stocks. Extreme allocation.

Black-Litterman: Express the view "stocks will outperform bonds by 2% more than equilibrium" with 60% confidence. The blended allocation might shift to 70/30 instead of 90/10. More moderate, more practical.

For the exam:

Be ready to:

• Describe both frameworks
• Identify when each is appropriate
• Explain why MVO is sensitive to inputs
• Explain how Black-Litterman blends views with equilibrium
• Recognise that Black-Litterman doesn't require full return vectors (just views)

Don't need to do the full Black-Litterman math — Level III tests it conceptually.

Real-world adoption:

Black-Litterman is the de facto standard at most large institutional asset managers (Goldman Sachs developed it; State Street, BlackRock, and others use variants). Pure MVO is mostly an academic baseline.