Sharpe, Information, and Sortino — The Three Amigos of Performance Measurement

Ever tried judging a movie by its popcorn sales? That’s what raw returns feel like. These three ratios are the ticket inspectors — they tell you whether the popcorn came with a plot twist or just salt.

Why this matters (and how this builds on what you already know)

You’ve already wrestled with translating policy into implementable portfolios and learned how to pick and monitor managers. You also saw performance attribution (where we blamed — and credited — individual bets). Now we ask: once a manager reports returns, how do we judge them fairly given risk and benchmark choice? That’s where Sharpe, Information, and Sortino come in. They turn raw returns into meaningful performance signals you can actually act on during manager selection and monitoring — and they help connect attribution outputs to real decisions, such as whether a manager’s edge is skill or just volatility dressed up in a tuxedo.

What is each ratio? (short, geek-friendly edition)

• Sharpe Ratio — Reward per unit of total risk. Compares portfolio excess return to its total volatility.
• Information Ratio (IR) — Active reward per unit of tracking risk (active risk). Compares portfolio active return vs. benchmark to the volatility of that active return.
• Sortino Ratio — Reward per unit of downside risk. Like Sharpe but only penalizes returns below a target (typically the risk-free rate or zero).

Formulas (put on your math glasses)

Sharpe = (Rp - Rf) / σp
Information Ratio = (Rp - Rb) / σ(Rp - Rb)
Sortino = (Rp - Rf) / DownsideDeviation

Where:

• Rp = portfolio return
• Rf = risk-free rate
• Rb = benchmark return
• σp = standard deviation of portfolio returns
• DownsideDeviation = standard deviation of returns below a target

How they differ and why you’d pick one over another

Intuition with analogies (because metaphors stick)

• Sharpe is like a car’s fuel efficiency (miles per gallon) — how far you go per unit consumed (risk). Great for comparing different cars (strategies) on the open highway (no benchmark).

• Information Ratio is like how well a driver improves on a GPS route — it measures the additional minutes saved per minute of route deviation. If the benchmark is the GPS, IR tells you whether the driver’s shortcuts are consistently better, or just random detours.

• Sortino is like judging a chef by how often they avoid burning the meal, not by how flamboyant the garnishes are. You're mainly worried about bad outcomes.

Concrete example (numbers so your spreadsheet stops crying)

Imagine annualized data:

• Portfolio return Rp = 12%
• Benchmark return Rb = 8%
• Risk-free Rf = 2%
• Portfolio σp = 10%
• Tracking error σ(Rp - Rb) = 4%
• Downside deviation = 6%

Calculations:

• Sharpe = (12% - 2%) / 10% = 1.0
• Information Ratio = (12% - 8%) / 4% = 1.0
• Sortino = (12% - 2%) / 6% ≈ 1.67

Interpretation: Good absolute risk-adjusted return (Sharpe 1), solid active skill vs benchmark (IR 1), and an even better profile if you only care about downside protection (Sortino 1.67).

Where these metrics plug into your toolkit

• Manager selection and monitoring: IR is your frontline metric for active managers — it aligns with manager selection and monitoring because it focuses on the value added relative to the benchmark. Use Sharpe as a sanity check when selecting across different mandates (equity vs balanced vs hedge funds).

• Performance attribution: If attribution shows alpha concentrated in a subset of positions, compare the IR to the contribution from those positions. High IR + concentrated attribution = true skill (maybe). Low IR + noisy attribution = luck.

• Implementation shortfall control: When monitoring execution cost, compare realized active returns net of implementation shortfall. If implementation shortfall is large enough to materially reduce IR, your “alpha” might evaporate. If a manager’s reported active return disappears after factoring transaction costs, don’t just blame slippage — revisit trade implementation and rebalancing policy.

Common mistakes (read this before you hit publish on that manager memo)

1. Comparing IRs across different benchmarks. Bad idea. Benchmark choice changes tracking error and magnitude of active return.
2. Using Sharpe for strategies with skewed returns. Hedge funds and options strategies often have low Sharpe but acceptable Sortino — upside volatility shouldn’t be punished equally.
3. Ignoring horizon and frequency. Monthly vs daily returns change σ and downside deviation. Be consistent.
4. Small sample illusions. A hot quarter can spike IR; statistical significance matters. Look at t-stats or information coefficient equivalents.
5. Net vs Gross returns. Use net-of-fees returns for manager selection and monitoring unless you explicitly want to evaluate pre-fee skill.

Quick code/pseudocode — how you’d compute these in a notebook

# assume returns array r, benchmark returns rb, risk_free rate rf, target = rf
excess = r - rf
active = r - rb
sharpe = mean(excess) / std(r)
ir = mean(active) / std(active)
downside = std([x for x in excess if x < 0])  # or use target
sortino = mean(excess) / downside

Pro tip: annualize means converting mean and std appropriately (multiply mean by periods, std by sqrt(periods)).

Quick decision rules (if you like the short menu)

• Use Information Ratio first when evaluating an active manager vs a known benchmark.
• Use Sharpe when comparing across unbenchmarked strategies or portfolios with very different mandates.
• Use Sortino when downside protection matters more than symmetric volatility (pension funds, liabilities, tail-risk constraints).

“A ratio without context is just numbers arguing. Put it back into your mandate, benchmark choice, and implementation reality.”

Final bite — Key takeaways

• Sharpe, Information, and Sortino each answer a different question: total risk-adjusted return, active skill per unit of tracking risk, and downside-adjusted return, respectively.
• They complement attribution and implementation controls. Attribution tells you where returns came from; these ratios tell you whether those returns are worth their risk and survive execution costs.
• Always align metric → mandate → decision. Don’t judge a liability-driven instrument by Sharpe alone. Don’t celebrate IR without confirming net-of-fee performance and realistic implementation slippage.

Go forth and measure like a merciless, kind-hearted quant. Judge managers with ratios, but also with context, persistence, and the occasional suspicious eyebrow.

Version: Sharpy TA — Ratios That Actually Help You Hire People