How should analysts handle non-normality (fat tails and skewness) in historical return data for CME?

This is a great question because the curriculum takes a nuanced stance — it acknowledges non-normality is real but argues that accounting for it isn't always worth the analytical cost.

The Reality of Return Distributions:

Historical asset returns consistently show:

• Negative skewness — large losses occur more frequently than a normal distribution predicts
• Excess kurtosis (fat tails) — extreme returns (both positive and negative) are more common than normal
• These features are more pronounced for equities, hedge funds, and credit-sensitive instruments

When It Matters:

Why It's Often Not Worth the Cost:

Modeling non-normal distributions requires:

1. Estimating additional parameters (skewness, kurtosis, or full distribution shape)
2. More complex optimization frameworks (no closed-form MVO solution)
3. Larger data samples to estimate higher moments reliably
4. Reduced transparency — stakeholders understand mean-variance; they may not understand the Cornish-Fisher expansion

For strategic asset allocation with a 10+ year horizon, the central limit theorem pushes multi-period returns toward normality anyway. The added complexity of modeling fat tails for long-horizon allocation decisions often produces marginal improvement in allocation quality.

When You MUST Account for Non-Normality:

• Tail risk measurement (VaR, CVaR, stress testing)
• Short-horizon tactical decisions where single-period tail events matter
• Alternative investments with highly asymmetric payoffs (e.g., options, distressed credit)
• Any analysis where the user specifically cares about downside or extreme outcomes

Practical Approach:

Use normal assumptions for strategic allocation as the baseline, then overlay tail-risk analysis separately. This keeps the core framework tractable while acknowledging that extreme events require dedicated attention.

Practice non-normality questions in our CFA Level III question bank.