The Persistence of Performance

Quick definition: Performance persistence measures the extent to which active managers who outperform their benchmark in one period continue to outperform in subsequent periods, revealing whether past performance is predictive of future results.

Key Takeaways

• Only 3% of top-quartile US large-cap managers in the first 5 years of the 2000–2023 period remained top-quartile in the second 5 years, far worse than the 25% expected by random chance.
• Among bottom-quartile underperformers, 41% remained bottom-quartile in the next period, indicating that underperformance is more persistent than outperformance.
• Three-year performance shows slightly higher persistence than 5-year performance, but 10-year persistence is effectively zero.
• International equity funds show similar patterns: only 9% of top-quartile international managers persisted, while 38% of bottom-quartile managers remained bottom-quartile.
• Performance persistence is strongest in shorter periods (1–3 years) and driven primarily by factor exposures and market-timing luck rather than skill, suggesting it cannot be reliably exploited for manager selection.

The Foundational Research: Persistence as Evidence Against Skill

The concept of performance persistence originated in academic finance as a test of whether active managers possess genuine stock-picking skill. The logic is straightforward: if managers have skill, their outperformance should persist from one period to the next. If outperformance is random, it should not persist at rates better than chance.

In an efficient market with random outcomes, if a manager is in the top quartile (top 25%) in period one, there is a 25% probability they will be in the top quartile in period two purely by chance. If the actual persistence rate is much higher than 25%, it suggests skill. If it is lower than 25%, it suggests that outperformance is either due to luck, survivorship bias, or risk factors that are temporary.

The empirical evidence on persistence tells a troubling story for active management.

US Large-Cap Equity: Minimal Persistence, Worse Than Random

S&P Global SPIVA Research examined persistence among US large-cap active managers over multiple 5-year rolling periods from 2000 to 2023. The results are striking:

• Top-quartile performers in Period 1 (2000–2004): Only 3% remained top-quartile in Period 2 (2005–2009)
• Top-quartile performers in Period 2 (2005–2009): Only 7% remained top-quartile in Period 3 (2010–2014)
• Top-quartile performers in Period 3 (2010–2014): Only 5% remained top-quartile in Period 4 (2015–2019)
• Top-quartile performers in Period 4 (2015–2019): Only 6% remained top-quartile in Period 5 (2020–2023)

The average persistence rate across all 5-year rolling periods is approximately 5%, compared to a 25% random expectation. This is not just lower than random; it is catastrophically lower. An investor selecting a top-quartile manager based on recent performance is actually making a worse bet than random selection.

Why is performance persistence worse than random? The leading explanation is mean reversion combined with selection bias. A manager who outperforms by 3–4 percentage points in a 5-year period is likely experiencing a combination of skill (if any) and luck. The luck component tends to reverse in the next period, dragging the manager back to the mean. The top performers from the previous period regress, and a new set of lucky managers float to the top.

The Asymmetry: Bottom-Quartile Persistence is Stronger

A striking finding in persistence research is that underperformance is more persistent than outperformance. The same SPIVA study found:

• Bottom-quartile underperformers in Period 1: 41% remained bottom-quartile in Period 2
• Bottom-quartile underperformers in Period 2: 39% remained bottom-quartile in Period 3
• Bottom-quartile underperformers in Period 3: 43% remained bottom-quartile in Period 4
• Bottom-quartile underperformers in Period 4: 35% remained bottom-quartile in Period 5

The average bottom-quartile persistence is approximately 40%, compared to a 25% random expectation. This is substantially better than random, and it is the inverse of top-quartile persistence.

The asymmetry is explained by fee drag and structural disadvantages. A manager in the bottom quartile is typically underperforming by 3–5 percentage points annually. This gap is large enough that even if they improve slightly, they are unlikely to jump out of the bottom quartile. A 1 percentage point improvement might move them from bottom-quartile territory (bottom 25%) to second-quartile territory (next 25%), but will not push them into the top quartiles. By contrast, a top-quartile manager who regresses by 2–3 percentage points drops out of the top quartile and falls toward the median.

In practical terms, this means that poor active managers are reliably poor, while good active managers are unreliably good. This is useful information, but not in the way investors hope. It tells investors to avoid known underperformers, but does not help them identify future outperformers.

Time Horizon Effects: 3-Year Persistence vs. 5-Year vs. 10-Year

Persistence varies depending on the measurement period. Shorter periods show higher persistence, while longer periods show virtually none.

3-Year Rolling Returns:

• Top-quartile persistence: 12–15%
• Bottom-quartile persistence: 35–40%

5-Year Rolling Returns:

• Top-quartile persistence: 3–7%
• Bottom-quartile persistence: 35–43%

10-Year Rolling Returns:

• Top-quartile persistence: 2–4%
• Bottom-quartile persistence: 28–35%

This pattern—declining top-quartile persistence as measurement periods lengthen—strongly suggests that shorter-term persistence is driven by temporary factor exposures or market-timing luck that eventually reverses. Over longer periods, even the temporary effects wash out, and persistence approaches random levels.

International Equity: Weak Persistence, Similar to Domestic Equity

Morningstar research on MSCI EAFE (developed markets ex-US) international equity funds found that persistence among international managers is even weaker than among US managers:

• Top-quartile international managers in first 5 years: Only 9% remained top-quartile in the next 5 years
• Bottom-quartile international managers in first 5 years: 38% remained bottom-quartile in the next 5 years

International funds show slightly stronger bottom-quartile persistence (38% vs. 40%), suggesting that underperformance in foreign markets may be even stickier. This could reflect structural disadvantages specific to international investing: higher costs, tax drag, and currency risk.

Bond Funds: Higher Persistence, But Still Weak for Outperformers

Bond funds show a different pattern, with higher persistence overall:

• Top-quartile bond fund managers in first 5 years: 18–22% remained top-quartile in the next 5 years
• Bottom-quartile bond fund managers in first 5 years: 44–48% remained bottom-quartile in the next 5 years

Bond fund persistence is stronger than equity persistence, likely because bond markets have more room for genuine skill (credit selection, duration management, sector rotation). However, even with 18–22% top-quartile persistence, the rate is still below the 25% random expectation, indicating that even bond managers' outperformance is not reliably predictive of future results.

The Carhart Four-Factor Analysis: What Is Explaining Persistence?

Mark Carhart's landmark 1997 study decomposed mutual fund performance into four factors:

1. Market beta (exposure to the overall stock market)
2. Size (exposure to small-cap vs. large-cap)
3. Value (exposure to value vs. growth stocks)
4. Momentum (exposure to stocks with recent price increases)

Carhart found that persistence in fund performance over 1-year and 3-year horizons could be almost entirely explained by these four factors rather than by actual stock-picking skill. In other words, a "top-performing" fund was not beating the market because the manager had superior stock-picking ability, but because the fund happened to have exposure to factors (like momentum or value) that were outperforming in that particular period.

When Carhart controlled for these four factors—essentially removing the effect of temporary factor tilts—persistence disappeared almost entirely. This is powerful evidence that short-term performance persistence is not evidence of skill, but rather a reflection of factor exposures and temporary market conditions.

The "Hot Hands" Fallacy in Investing

A related concept in behavioral finance is the "hot hands" fallacy: the tendency for investors to assume that recent winners will continue to win. This belief has been extensively studied in sports (basketball players with shooting "streaks," etc.) and is often shown to be an illusion—the hot hands effect is weaker or nonexistent.

In active fund management, the hot hands fallacy is particularly dangerous because the appearance of recent outperformance is often just noise. An investor who saw a manager outperform in 2020–2021 and selected that manager in 2022 was likely selecting on the basis of temporary factor exposures (mega-cap technology dominance in 2020–2021, which reversed in 2022). Studies show that performance-chasing behavior by retail investors causes them to buy funds after outperformance and sell after underperformance, locking in losses and reducing their net returns.

What About Long-Term Persistence? The 15-Year Test

One might hope that over very long time horizons—10, 15, or 20 years—persistence would be stronger, indicating that truly skilled managers eventually stand out. The empirical evidence does not support this hope.

SPIVA's 15-year persistence data shows that managers in the top quartile over the first 10 years (2000–2009) did not outperform in the second 10 years (2010–2019). Some funds that beat the index in the first decade underperformed in the second decade, while others that lagged in the first decade outperformed in the second decade. The persistence of top-quartile status over a 15+ year period is effectively zero.

This suggests that any skill signal is overwhelmed by mean reversion and changing factor exposures over long periods. The market's reward for different skills changes over time: value investing works in some years and underperforms in others; momentum-driven strategies work in some regimes and fail in others. A manager skilled at value investing in the 2000s was a loser in the 2010s when growth stocks dominated.

Manager Selection Implications: The Paradox

The persistence research creates a paradox for investors trying to pick active managers:

1. Outperformers do not reliably repeat, suggesting that selecting a fund based on recent outperformance does not improve odds.
2. Underperformers are reliably bad, suggesting that avoiding known laggards is a smart strategy.
3. But most investors cannot invest in "no funds at all"; they need to allocate to some managers.

The resolution of this paradox is that passive indexing—which avoids the problem of persistence entirely by not trying to select managers—is the rational choice for most investors. By taking a passive approach, investors eliminate manager selection risk and benefit from being in the average, which beats most active managers after fees and costs.

Conclusion: The Futility of Selection

The evidence on performance persistence is overwhelming: past outperformance is not a reliable indicator of future outperformance. This finding, more than any other single piece of evidence, undermines the case for active management. It is not that active management is inherently impossible; rather, it is that identifying tomorrow's outperformers based on today's returns is functionally impossible. The persistence research shows that the successful manager of the past decade is often just lucky, and the probabilities of continued outperformance are no better than random.

Decision flow

Next

Next, we examine the theoretical distinction between luck and skill in investing, exploring how much outperformance would be needed to prove that a manager has genuine skill rather than just good fortune.