Module 4 — Portfolio Construction

Sharpe, Sortino, and Information Ratio

From Markowitz to drawdown-aware sizing. The math behind every allocator's spreadsheet.

Learning objectives

  • Compute Sharpe correctly (annualised, with the right risk-free rate).
  • Understand when Sortino or IR is more appropriate.
  • Recognize common Sharpe-inflation tricks.

FORMULA

Sharpe ratio

SR = (E[r_p] - r_f) / σ_p
(use annualised numerator and denominator)

TEXT

Sharpe rules of thumb

Long-only equities: ~0.4 long-term. Good single-name long/short: ~0.7–1.0. Mid-frequency systematic strategy: ~1.5–2.0 net of costs. HFT market-making: 5+ but with very low capacity. Anything reporting Sharpe > 3 in a backtest is almost certainly overfit or has a bug.

FORMULA

Sortino (downside-only)

Sortino = (E[r_p] - target) / σ_downside
where σ_downside uses only returns below target. Useful when upside volatility shouldn't be penalised.

FORMULA

Information ratio

IR = (E[r_p - r_bench]) / σ(r_p - r_bench)
Measures alpha generation per unit of tracking error. The benchmark version of Sharpe.

TEXT

Common ways to inflate Sharpe

• Use total return but daily-frequency T-bill rate (mismatched). • Ignore transaction costs and borrow fees. • Backtest on a survivor-only universe. • Pick the start date that omits a drawdown. • Compute on overlapping windows (autocorrelation deflates σ).