Quantitative Methods and Statistics Flashcards

The Calmar ratio is calculated as:

Answer: Annualized return divided by maximum drawdown

The Calmar ratio equals the compound annualized return divided by the absolute value of the maximum drawdown, making it a useful risk-adjusted measure for strategies where large peak-to-trough losses are the primary concern.

When computing the correlation between a liquid equity index and an illiquid private asset, which adjustment is most important?

Answer: Autocorrelation (stale pricing) adjustment to unsmooth the illiquid asset returns

Stale pricing in illiquid assets induces artificial autocorrelation that suppresses measured volatility and correlation; unsmoothing the return series (e.g., Geltner adjustment) yields more economically accurate correlation estimates.

How does cross-sectional momentum differ from time-series (absolute) momentum?

Answer: Time-series momentum compares an asset's own recent return to its historical average; cross-sectional momentum ranks assets relative to each other at a point in time

Time-series momentum goes long an asset when its own recent return is positive (absolute trend), while cross-sectional momentum goes long the top-performing assets and short the worst relative to each other at a given date.

The Brinson-Hood-Beebower (BHB) performance attribution model decomposes active return into:

Answer: Allocation effect, selection effect, and interaction effect

BHB attribution separates the manager's active return into the allocation effect (overweighting/underweighting segments), the selection effect (picking better securities within segments), and their interaction.

Look-ahead bias in backtesting occurs when:

Answer: The model inadvertently incorporates information that would not have been available at the time of the simulated trade

Look-ahead bias contaminates backtests when data unavailable to a real investor at the decision date (e.g., revised earnings, end-of-month prices) are used as inputs, producing artificially inflated simulated returns.

For which type of alternative investment return distribution does the law of large numbers converge most slowly, making statistical inference most difficult?

Answer: Returns from strategies with fat tails and non-stationary parameters

The law of large numbers converges much more slowly for heavy-tailed (high kurtosis) and non-stationary distributions because rare extreme events dominate the sample statistics, requiring very large sample sizes for reliable inference.

Which of the following best describes the concept of 'variance drag' (also called volatility drag) in compounded returns?

Answer: The difference between the arithmetic mean return and the lower compounded (geometric) return caused by return volatility

Variance drag quantifies how volatility reduces compounded wealth growth; the geometric mean ≈ arithmetic mean − (σ²/2), so higher volatility causes compounded returns to fall further below the simple average.