The Information Ratio (IR) divides active return (alpha) by tracking error: IR = (Rportfolio − Rbenchmark) / TE. It measures how much active return a manager generates for each unit of active risk taken. IR > 0.5 = consistently adding value; IR > 1.0 = exceptional (very rare); IR < 0 = subtracting value vs the benchmark. For passive index trackers, IR is meaningless — they aren't trying to add active return. IR is relevant for active ETFs, smart-beta ETFs, and any strategy claiming to outperform a benchmark.
The formula
IR = (Rp − Rb) / TE where: Rp = annualised return of the active portfolio/ETF Rb = annualised return of the benchmark TE = annualised tracking error (std dev of Rp − Rb) The numerator is "alpha" or "active return" — what the manager added above the benchmark. The denominator is tracking error — the variability of that active return. The ratio answers: per unit of active risk, how much active return was delivered?
IR vs Sharpe — the conceptual difference
| Sharpe ratio | Information ratio | |
|---|---|---|
| Numerator | Excess return over risk-free rate | Excess return over benchmark |
| Denominator | Total volatility (σ) | Tracking error (TE) |
| Best for evaluating | Absolute return strategies | Active/relative-return strategies |
| Use for passive trackers | Yes — meaningful | No — undefined / not relevant |
| Use for active funds | Reasonable | More appropriate |
If you're measuring "did this strategy beat its benchmark?" — use IR. If you're measuring "did this strategy beat cash?" — use Sharpe.
Inputs we use
| Input | Source | Notes |
|---|---|---|
| Portfolio monthly returns | Issuer + LSE | 36 months, GBP TR |
| Benchmark monthly returns | Stated benchmark from issuer factsheet | 36 months, GBP TR |
| Active returns (Di) | Rp,i − Rb,i for each month | Sign matters — positive = outperformance |
| Annualisation | Mean scaling for active return; √12 for TE | Same as Sharpe |
Worked example — hypothetical active UK equity ETF
Consider an active UK equity ETF with the FTSE All-Share as benchmark, 3 years ending April 2026.
| Active ETF annualised return | +11.5% |
| FTSE All-Share annualised return | +9.8% |
| Active return (alpha) = 11.5 − 9.8 | +1.7% |
| Monthly return differences over 36 months | ... |
| Standard deviation of monthly differences | ~0.95% |
| Annualised tracking error = 0.95% × √12 | ~3.3% |
| Information Ratio = 1.7 / 3.3 | ~0.52 |
An IR of 0.52 over 3 years is reasonable — the manager added 1.7% per year of active return at the cost of 3.3% per year of tracking error. By academic standards (Grinold 1989), an IR of 0.5 is "good"; 0.75 is "very good"; 1.0+ is "exceptional" and rare.
Interpreting IR levels
| IR range | Interpretation | How common in 3-year windows |
|---|---|---|
| IR < 0 | Active management subtracted value vs benchmark | ~50% of active funds |
| 0 to 0.25 | Marginal value-add, not enough for fees | ~25% of active funds |
| 0.25 to 0.50 | Some skill, but may not justify fees vs index alternative | ~15% of active funds |
| 0.50 to 0.75 | Good — consistent alpha generation | ~7% of active funds |
| 0.75 to 1.00 | Very good — clear value-add | ~2% of active funds |
| 1.00+ | Exceptional — rare and often hard to sustain | ~1% of active funds |
The market is roughly zero-sum (one manager's alpha is another's loss). Mathematically, most active managers must have IR near zero. Active managers with persistent IR > 0.5 over decades are very rare — and that's why they're famous (and expensive).
The IR-statistical-significance question
Important nuance: a 3-year IR of 0.5 is exciting but not statistically meaningful. To know whether observed alpha represents real skill or luck:
- Approximate t-stat: t ≈ IR × √(years of data).
- Threshold for statistical significance (95%): t > 2.0.
- So IR 0.5 over 3 years: t ≈ 0.87 — not significant.
- IR 0.5 needs ~16 years to clear the 95% confidence bar.
- IR 1.0 needs ~4 years.
Implication: most "evidence" of active manager skill from 3-5 year track records is statistically noise. Long-term IR > 0.5 is the real benchmark for proving skill.
Alpha — the numerator of IR explained
"Alpha" in the IR formula is the excess return above the benchmark. But "alpha" has a specific academic meaning in CAPM:
CAPM alpha = Rportfolio − [Rf + β × (Rbenchmark − Rf)] CAPM alpha adjusts for beta — i.e. credits the portfolio only for return above what was expected given its market sensitivity.
For most UK retail purposes, the simpler "active return = ETF − benchmark" is sufficient. CAPM alpha gets used in academic papers and institutional fund evaluation.
What IR does NOT tell you
- Future performance. Past IR is descriptive, not predictive.
- Cost-adjusted performance. IR uses post-OCF returns; some sources publish pre-OCF IR (much higher numbers — be careful).
- Risk-of-loss. A manager with IR 0.5 can still have years of significant underperformance — IR averages it out.
- Skill vs luck. Without statistical-significance analysis, IR can be flattered by short-period luck.
- Whether the active risk was sensible. IR 0.7 with TE 4% is impressive; IR 0.7 with TE 12% means the manager took on huge active risk for the alpha generated.
How to reproduce this yourself
- Get 36 months of ETF returns (GBP TR).
- Get 36 months of benchmark returns (same currency, total return).
- Calculate monthly active returns: RETF − Rbenchmark.
- Calculate annualised mean active return: AVERAGE × 12 (or geometric: =((1+mean)^12)-1).
- Calculate annualised tracking error: STDEV.S(differences) × SQRT(12).
- IR = annualised active return / annualised tracking error.
Sources and methodology
Information ratio originally defined by Treynor & Black (1973); formalised by Grinold (1989). Standard institutional practice (Grinold & Kahn, Active Portfolio Management). The ETF Data Methodology documents all data sources. The site methodology documents the broader review process.
Related metric pages
How UK Tax Drag holds itself to account
Every page is reviewed against the editorial standards, written from primary sources, sourced openly, and corrected publicly. No affiliate revenue. No sponsored content. No paid placements.