CH01.05
Sampling Uncertainty of a Sharpe Ratio
QR · PM · RM · difficulty 2/5 · 15 minutes
Interview question
A daily strategy reports an annualized Sharpe ratio of 1.5 from three
years of data. Estimate its uncertainty under IID Gaussian returns.
Then explain how skewness, kurtosis and autocorrelation change the answer.
30-second answer
With 756 daily observations, the IID Gaussian asymptotic standard
error of the annualized Sharpe is about 0.58, giving a rough 95%
interval of 0.37 to 2.63. The reported 1.5 is much less precise
than it looks.
Full solution — Lo (2002); Mertens (2002)
s = 1.5 / √252 ≈ 0.09449
(1.42)
SE(ŜA) ≈ √252 · ((1 + ½s²) / n)1/2 ≈ 0.579
(1.45)
CI95% ≈ [0.37, 2.63]
(1.47)
neff ≈ n (1 − ρ₁) / (1 + ρ₁)
(1.49)
Lab 1.5 — Implementation contract
| Input | annualized Sharpe, observations, periods per year, skewness, excess kurtosis |
| Invariant | SE scales as inverse square root of n; rises under adverse higher moments |
| Failure | n ≤ 1, or the moment combination implies negative variance |
Unit verification
test_annualized_sharpe_standard_error_matches_reference_case ✓ 0.578
test_annualized_sharpe_se_scales_and_reflects_higher_moments ✓ higher moments covered