Information Ratio: Measuring Active Manager Skill
The Information Ratio is the definitive metric for evaluating active investment managers. Unlike the Sharpe Ratio, which measures total return per unit of total risk, the IR isolates the skill component — how consistently a manager generates returns above their benchmark, relative to how much they deviate from it.
1. The Formula
The Information Ratio is defined as:
IR = (Rp - Rb) / σ(Rp - Rb)
Where:
Rp= portfolio returnRb= benchmark returnRp - Rb= active return (the excess return generated by active decisions)σ(Rp - Rb)= tracking error (the standard deviation of active returns)
The numerator, active return, measures how much value the manager added (or subtracted) relative to a passive benchmark. The denominator, tracking error, measures how much risk the manager took to generate that active return. A manager who beats the benchmark by 2% per year with a tracking error of 4% has an IR of 0.5. Another manager who beats the benchmark by 3% per year with a tracking error of 12% has an IR of only 0.25 — less consistent despite higher raw outperformance.
2. Active Return: The Numerator
Active return is simply the difference between the portfolio’s return and the benchmark’s return over the same period. If your portfolio returned 12% and the S&P 500 returned 10%, your active return is +2%. If your portfolio returned 8% and the benchmark returned 10%, your active return is −2%.
It is essential that the benchmark is appropriate. An emerging markets equity fund should be measured against an emerging markets index (such as the MSCI Emerging Markets Index), not the S&P 500. A small-cap value fund should be measured against a small-cap value benchmark. Using the wrong benchmark makes the IR meaningless — a small-cap fund will appear to have enormous active returns (both positive and negative) when measured against a large-cap index, but this is driven by factor exposures, not stock selection skill.
Arithmetic vs. Geometric Active Return
Active return can be computed arithmetically (simple difference) or geometrically (ratio minus one). For short periods (monthly, quarterly), the two are nearly identical. Over longer periods, the geometric calculation is more accurate because it correctly accounts for compounding. The geometric active return for a single period is:
Geometric active return = (1 + Rp) / (1 + Rb) - 1
In practice, most IR calculations use arithmetic active returns computed at the monthly level and then annualize.
3. Tracking Error: The Denominator
Tracking error (also called active risk) is the standard deviation of active returns. It measures the variability of the manager’s performance relative to the benchmark. A low tracking error means the portfolio closely follows the benchmark; a high tracking error means the manager is making large active bets that cause performance to deviate significantly from the index.
Typical tracking error ranges for different fund types:
| Fund Type | Typical Tracking Error |
|---|---|
| Enhanced index / closet indexer | 0.5% – 2% |
| Traditional active large-cap | 3% – 6% |
| Concentrated stock picker | 6% – 12% |
| Sector / thematic fund | 8% – 15%+ |
| Long/short equity hedge fund | 5% – 15% |
A tracking error below 1% is often a sign of a “closet indexer” — a fund that charges active management fees but largely replicates the benchmark. These funds almost by definition will have low active returns as well, and the fees they charge often exceed their active return, resulting in a negative IR.
4. Interpreting the Information Ratio
What constitutes a “good” Information Ratio? The conventional benchmarks, widely used in the institutional investment industry, are:
| IR Range | Interpretation |
|---|---|
| Below 0 | Manager underperforms the benchmark after adjusting for active risk |
| 0.0 – 0.25 | Below average — modest or inconsistent outperformance |
| 0.25 – 0.50 | Average to above average for an active manager |
| 0.50 – 0.75 | Good — consistent, meaningful outperformance |
| 0.75 – 1.00 | Very good — rare among traditional active managers |
| Above 1.0 | Exceptional — very rare on a sustained basis |
Sustaining an IR above 0.5 over a full market cycle (typically 7 to 10 years) is genuinely difficult. An IR above 1.0 sustained over a decade is almost unheard of in traditional long-only equity management. Some quantitative strategies and hedge funds have achieved it, but usually with significant constraints on capacity.
An IR of 0.5 in large-cap U.S. equities (one of the most efficient markets in the world) represents far more skill than an IR of 0.5 in frontier markets or small-cap emerging markets, where information asymmetries and structural inefficiencies are much larger. Always consider the opportunity set when comparing IRs across managers in different asset classes.
5. Information Ratio vs. Sharpe Ratio
The Sharpe Ratio and the Information Ratio are closely related but answer fundamentally different questions. The Sharpe Ratio measures total excess return (above the risk-free rate) per unit of total risk (total standard deviation of returns):
Sharpe = (Rp - Rf) / σ(Rp)
The Information Ratio measures active return (above the benchmark) per unit of active risk (tracking error):
IR = (Rp - Rb) / σ(Rp - Rb)
The critical difference is in the reference point. The Sharpe Ratio uses the risk-free rate; the IR uses a benchmark. This distinction has important consequences:
- A leveraged index fund can have a high Sharpe Ratio (because it amplifies both return and risk roughly equally, and the numerator grows with leverage while the denominator is total vol). But it has an IR of zero (or slightly negative due to borrowing costs) because it generates no active return.
- A cash-plus strategy (absolute return target) has a Sharpe Ratio that equals its IR when the benchmark is set to the risk-free rate. In this special case, the two metrics are identical.
- A long-only equity fund can have a low Sharpe Ratio (because equities are volatile) but a good IR (because it consistently beats its benchmark). The Sharpe Ratio captures market risk; the IR isolates manager skill.
When evaluating an active manager, the IR is almost always the more relevant metric. The Sharpe Ratio conflates the market return (which requires no skill — just buy an index fund) with the active return (which reflects the manager’s stock selection and timing decisions).
6. Grinold’s Fundamental Law of Active Management
In 1989, Richard Grinold published a remarkable result that connects the Information Ratio to two intuitive components of investment skill. The Fundamental Law of Active Management, first presented in his paper “The Fundamental Law of Active Management” in the Journal of Portfolio Management, Vol. 15, No. 3, states:
IR ≈ IC × √BR
Where:
IC= Information Coefficient, the correlation between the manager’s predicted returns and the actual returns that materialize. This measures forecasting skill. An IC of 0.05 means the manager’s forecasts have a 5% correlation with realized returns — which sounds low but is actually quite good for equity markets.BR= Breadth, the number of independent investment bets (independent forecasts) the manager makes per year.
The law says that the IR is the product of skill (IC) and the square root of the number of independent opportunities (BR). This has profound implications for how different investment styles generate alpha.
The Concentrated Stock Picker
A concentrated stock picker holds 20 to 30 positions, rebalances quarterly, and makes perhaps 20 truly independent investment decisions per year (BR = 20). To achieve an IR of 0.5, they need an IC of 0.5 / √20 ≈ 0.11 — an exceptionally high forecasting accuracy. This is the Warren Buffett path: make a small number of bets, but be right a disproportionate amount of the time.
The Quantitative Strategist
A quantitative equity strategy might trade 500 stocks with weekly rebalancing, making roughly 500 × 52 / 2 ≈ 13,000 position-level decisions per year. If even half of those are truly independent, BR ≈ 6,500 and √BR ≈ 80.6. To achieve the same IR of 0.5, the quant needs an IC of only 0.5 / 80.6 ≈ 0.006 — a tiny forecasting edge per individual bet. This is the Renaissance Technologies path: make thousands of small bets with a barely perceptible edge on each one.
The Fundamental Law explains why both approaches can work. The concentrated picker needs high IC (hard to achieve, hard to sustain). The quant needs high breadth (requires infrastructure, data, and models, but the per-bet skill requirement is much lower). It also explains why quant strategies can be more consistent: the Law of Large Numbers smooths out the variance across many small bets.
The Fundamental Law assumes that all bets are independent, which is rarely true in practice. Correlated bets reduce the effective breadth. If you hold 100 tech stocks that all move together, your effective breadth might be closer to 10 than 100. Clarke, de Silva, and Thorley extended the law in 2002 to account for a “transfer coefficient” (TC) that captures constraints like long-only mandates: IR ≈ IC × TC × √BR.
7. Annualizing the Information Ratio
The IR is typically reported on an annualized basis. If you compute the IR from monthly data, the annualization formula is:
Annualized IR = Monthly IR × √12
This works because the active return scales linearly with time (multiply monthly active return by 12) and the tracking error scales with the square root of time (multiply monthly tracking error by √12). The ratio of the two gives a √12 scaling factor.
For quarterly data:
Annualized IR = Quarterly IR × √4 = Quarterly IR × 2
And for daily data:
Annualized IR = Daily IR × √252
Where 252 is the approximate number of trading days in a year.
8. Statistical Significance and Sample Size
One of the most important — and most overlooked — aspects of the IR is its sampling uncertainty. The IR is estimated from a finite sample of returns, and like any sample statistic, it has a standard error. The approximate standard error of the IR estimated from N independent return observations is:
SE(IR) ≈ √((1 + IR² / 2) / N)
For an IR of 0.5 estimated from 60 monthly observations (5 years), the standard error is approximately √((1 + 0.125) / 60) ≈ 0.137. A 95% confidence interval is roughly 0.5 ± 0.27, spanning from 0.23 to 0.77. This is an enormous range. After 5 years of data, you still cannot distinguish with high confidence between a mediocre manager (IR = 0.23) and an excellent one (IR = 0.77).
This is why the institutional investment community demands long track records. Even 10 years of monthly data (120 observations) only narrows the standard error to about 0.097, giving a 95% confidence interval of roughly 0.5 ± 0.19. The implication is sobering: you need decades of data to be statistically confident that a manager has genuine skill, by which time the manager, the market, or the strategy may have fundamentally changed.
Be extremely cautious with IR estimates based on short track records. A manager with a 1.0 IR over 3 years could easily be a 0.3 IR manager who got lucky (or a 1.5 IR manager who got unlucky). The signal-to-noise ratio of performance data is far lower than most investors appreciate.
9. Practical Applications
Comparing Managers Against the Same Benchmark
The IR is most useful when comparing multiple managers who invest against the same benchmark. If three large-cap U.S. equity managers all benchmark to the S&P 500, the one with the highest IR over a full market cycle is, statistically, the most skilled (with the caveats about sample size above).
Setting Tracking Error Budgets
Institutional investors often set tracking error targets for their managers. If a plan sponsor wants 2% active return and believes its managers can sustain an IR of 0.4, it should budget a tracking error of 2% / 0.4 = 5%. This framework connects the performance target to the risk budget in a disciplined way.
Evaluating Strategy Changes
When a manager proposes to change their strategy (e.g., expanding into international stocks, adding factor tilts), the IR framework allows the plan sponsor to evaluate whether the new strategy is likely to improve the risk-adjusted active return. If the change increases breadth without diluting IC, the Fundamental Law predicts an improvement in IR.
10. Limitations of the Information Ratio
Like all performance metrics, the IR has limitations that practitioners should understand:
- Assumes normal distribution of active returns. If active returns are skewed (as they are for many hedge fund strategies), the IR can be misleading. A strategy that earns small positive active returns most of the time but occasionally suffers massive underperformance will have a deceptively high IR until the tail event occurs.
- Time-period dependent. The IR varies significantly depending on the measurement window. A manager with a high IR in bull markets may have a low or negative IR in bear markets. Always evaluate over a full cycle.
- Benchmark sensitivity. The IR changes if you change the benchmark. A manager could have an IR of 0.6 against the S&P 500 and an IR of 0.2 against the Russell 1000 Growth index, simply because their style tilts align more closely with one benchmark than the other.
- Does not capture tail risk. Two managers with the same IR can have very different maximum drawdown profiles. Supplement the IR with drawdown analysis and conditional performance metrics.
- Survivorship bias. Published IR data often suffers from survivorship bias — poor-performing funds are closed or merged, and their track records disappear from the databases, inflating the apparent average IR of surviving managers.
11. The IR in Context: Building a Complete Picture
The Information Ratio should never be used in isolation. A comprehensive evaluation of an active manager combines the IR with several complementary analyses: attribution analysis (where does the active return come from — sector bets, stock selection, factor tilts?), drawdown analysis (what is the worst-case underperformance?), hit rate (what fraction of bets are profitable?), and qualitative assessment of the investment process, team stability, and organizational alignment.
Together, these tools provide a much richer picture than any single metric. The IR tells you how efficiently a manager converts active risk into active return. Attribution tells you whether the sources of return are sustainable and intentional. Drawdown analysis tells you whether the ride is tolerable. And qualitative assessment tells you whether there is a coherent, repeatable process behind the numbers.
Active management is a difficult game. Most managers underperform their benchmarks over long periods after fees. But for those who can sustain a positive IR — particularly an IR above 0.5 — the compounding effect of consistent outperformance is substantial. The Information Ratio is the best single metric we have for identifying those rare managers.