How Long Would an Active Fund Manager Need to Demonstrate Outperformance to Be Confident in Their Results?
I have talked a lot about traditional discretionary active fund management and highlighted that the data clearly show that the vast majority consistently underperform a style-matched benchmark over any meaningful time period (ten years or more). This result holds across asset classes, geographies and macro-economic environments. However, the data are also clear that a small minority of active managers do appear to outperform, even over longer time horizons. So, to answer the question, how long of a time horizon would an investor need to be sure that the manager’s outperformance was down to skill and not luck, I conduct an analysis to quantify this seemingly subjective puzzle.
One of the key metrics when assessing whether an active manager is genuinely skilled (rather than simply lucky) is the Information ratio (IR). The IR measures how much excess return (alpha) a manager delivers per unit of active risk (tracking error). Mathematically:
IR = Average active return / Tracking error
Here, the average active return represents the mean of the manager’s returns above the benchmark, whilst the tracking error is the standard deviation of those excess returns.
For example, if a manager beats their benchmark by 1% per year on average and the volatility of that excess return is 2%, then:
IR = 1 / 2 = 0.5
Why Does This Metric Matter?
Even if a manager shows positive excess returns, we must ask: Could this simply be luck? To assess whether the observed outperformance is statistically different from zero, that is, evidence of genuine skill, we can approximate the t-statistic of the alpha estimate as:
t ≈ IR × √T
where T is the number of years of performance data.
A common rule of thumb for statistical significance is t ≈ 2, which corresponds roughly to 95% confidence that the observed alpha is not the result of chance. Rearranging the formula gives an estimate of how many years of data would be required:
T ≈ (t / IR)²
Applying the Numbers
If the manager’s information ratio is 0.5 and we set t = 2, then:
T = (2 / 0.5)² = 16
In other words, you would need about 16 years of data before you could be 95% confident that the manager’s outperformance reflected skill rather than luck.
However, if these IR numbers differed from 0.5, for example:
IR = 1.0 → T = (2 / 1.0)² = 4 years
IR = 0.25 → T = (2 / 0.25)² = 64 years
IR = 0.4 → T = (2 / 0.4)² = 25 years
IR = 0.2 → T = (2 / 0.2)² = 100 years
What Does This Tell Us?
An IR of 0.5 is actually quite strong by industry standards, yet it still implies that investors would need to observe around 16 years of consistent results to be confident that the manager is genuinely skilled. Most managers have lower IRs, often in the 0.2–0.4 range, which means it could take several decades of data to reach the same statistical confidence.
This illustrates why so few managers can reliably demonstrate persistent skill. Over short timeframes, luck dominates outcomes, making it extraordinarily difficult to separate true ability from random variation.
To believe a manager is skilled, you need both a high information ratio and a long, stable record of performance. For most discretionary active funds, the mathematics simply does not support the claim of skill.
