×
×
×
×
×

Tell us once and we'll remember.

I'm an...

Don't worry, you can always change this selection using the icons at the top left of the site.
  

The Hidden Risk of Crowding: Persistent Underperformance of Consensus Trades

October 2026

Key Takeaways
  • The most-crowded stocks have shown persistent benchmark-relative underperformance, making crowding diagnostics useful for identifying consensus-heavy positions where expectations and ownership may be stretched.
  • Crowded names have been most vulnerable during weak markets, liquidity stress and active equity outflow periods, so monitoring portfolio-level crowding can help reduce hidden concentration risk and improve drawdown control.
  • Less-crowded stocks have offered stronger relative return potential over time, suggesting active managers may add value by combining crowding data with bottom-up research to identify overlooked companies.

In 2019, we demonstrated that active crowded securities — those stocks for which sentiment has been broadly bullish among active investors for an extended period — were more likely to become overvalued. We examined active crowding as a risk factor and found that active uncrowded securities tended to outperform both active crowded securities and the broader universe over time. In addition, the return characteristics of active uncrowded portfolios suggested the existence of a meaningful universe of deeply undervalued securities.

Here we refine how crowding is measured by constructing the ClearBridge Active Crowding factor, bring our back test up to date through June 2026, and test the factor’s robustness across both U.S. and global equity markets.

The updated results reinforce crowding’s value as a portfolio risk lens.

Executive Summary of Results

Crowding risk is concentrated in the most-crowded stocks. Across the U.S. investable universe from January 2005 to June 2026, the most-crowded quintile consistently underperformed the Russell 3000 across multiple holding periods, with annualized benchmark-relative returns of roughly –3.0% to –3.5% equal-weighted and –3.5% to –4.0% value-weighted over one- to 36-month horizons.

The long-short spread is economically meaningful, though not always statistically significant monthly. Least-crowded portfolios generated meaningful cumulative return advantages over time, with the annualized return differential peaking at approximately 4% value-weighted and 5% equal-weighted for the twelve-month holding period.

Crowding risk is most visible in weak markets and stress periods. In months when the Russell 3000 return is below 0%, the least-crowded minus most-crowded portfolio earns about 1.5% per month, with a t-statistic of 2.9. The least-crowded quintile also outperformed during weak or low-return periods including 2008, 2011, 2015, 2018, 2022, March 2020 and the first quarter of 2026.

Market regime matters. Crowded stocks can outperform during strong risk-on or thematically driven periods, such as 2017, 2020, 2024 and 2025, when mega cap technology, pandemic digitization, AI infrastructure and other momentum themes supported heavily owned names. This reinforces the need to treat crowding as a risk lens rather than a mechanical sell signal.

Stock-level and factor evidence support the risk interpretation. On a 12-month basis, 56.7% of stocks in the most-crowded quintile underperformed the Russell 3000, with median underperformers lagging by –29.2% , while median outperformers exceeded it by +24.8%. The Active Crowding factor also adds incremental information beyond standard Fama-French factor exposures and tilts toward value, profitability and investment while leaning away from momentum.

Global results are broadly consistent with U.S. findings. In global universes, the most-crowded quintile underperformed the MSCI ACWI IMI benchmark by roughly –3.0% to –3.5% across 1- to 36-month horizons, with value-weighted returns reaching about –3.62% at the 12-month horizon.

Constructing the Active Crowding Factor

The ClearBridge Active Crowding factor combines four signals designed to capture different channels of crowded positioning: holding duration, active manager overweights, sell-side buy ratings and long-term price momentum (Exhibit 1).

The first two signals measure positioning: whether a stock is held for shorter periods by active managers and whether many active managers hold large benchmark-relative overweight positions. The other two signals measure sentiment: broadly positive sell-side ratings and strong trailing price momentum. Together, the four inputs are intended to identify stocks where ownership, sentiment and price action have become unusually aligned.

Exhibit 1: Components of the ClearBridge Active Crowding Factor

Exhibit 1: Components of the ClearBridge Active Crowding Factor

Source: ClearBridge Investments. Long-term price momentum is measured over the trailing two-year period (T−25 to T−2 months); sell-side buy ratings include ratings classified as buy or overbuy over the trailing three months.

¹Jain, K., & Jiao, D. (2025). Exploiting Myopia: The Returns to Long-Term Investing. SSRN

Constructing the Investable Universe

We test the factor on a U.S. institutional investable universe. The universe includes common stocks, REITs and MLPs listed on U.S. exchanges and excludes ADRs. We screen out stocks with trailing three-month average daily volume below the 25th percentile and market capitalization below the 20th percentile of the Russell 3000 Index. As of June 30, 2026, those thresholds translate to minimum average daily value traded of about $5.8 million and minimum market capitalization of about $560 million.

Using these liquidity and size filters helps ensure the results reflect stocks in which an active institutional equity portfolio could realistically take meaningful positions.

We convert each subfactor into a standardized z-score, combine the z-scores using a weighted average and assign each stock to a crowding quintile from 1 to 5. Q1 represents the least crowded stocks and Q5 represents the most crowded. We then back-test monthly rebalanced portfolios from January 2005 through June 2026.

Factor Performance Results: U.S.

To evaluate each input and the composite factor, we examine the skewness of subsequent one-month benchmark-relative returns across crowding quintiles. Across the four subfactors, the most-crowded quintiles generally show more negative return skew than the least-crowded quintiles, indicating greater downside asymmetry. The composite signal strengthens this pattern, suggesting that combining positioning and sentiment inputs provides a more complete view of crowding risk than any single component alone (Exhibit 2).

Exhibit 2: Results of ClearBridge Active Crowding Factor Test with Subfactors (U.S. Stocks)

Exhibit 2: Results of ClearBridge Active Crowding Factor Test with Subfactors (U.S. Stocks)

Source: ClearBridge Investments. January 2005 to June 2026. The analysis is conducted on U.S. stocks within the investable universe. Q1 = least crowded and Q5 = most crowded

Interpreting the U.S. Results
  • The annualized long–short return profile is cyclical rather than monotonic (Exhibit 3): leadership rotates between least-crowded and most-crowded portfolios across calendar years. That pattern suggests the factor is capturing shifts in positioning pressure and market appetite for consensus trades, not a simple rule that uncrowded stocks outperform in every environment.
  • Flows help explain when crowding helps or hurts. Strong inflows into active equities from 2004 to 2007 supported crowded names through continued buying pressure, while the outflow period from 2008 to 2011 exposed crowded portfolios to selling pressure and made less-crowded companies more resilient.
  • Crowded stocks can work when fundamentals and themes reinforce positioning. The strongest years for crowded names were not random; they coincided with broad macro or thematic tailwinds, including synchronized global growth in 2017, pandemic-era digitization in 2020 and AI infrastructure leadership in 2024.
    • 2017: Synchronized global growth and strong risk appetite fueled gains in mega cap tech names like Facebook, Apple, Amazon.com, Netflix, Google (now Alphabet), Microsoft and Nvidia — all heavily owned across active portfolios.
    • 2020: Pandemic-era digitization trends drove crowded stocks like Twilio, MercadoLibre, Chegg, Carvana, BJ’s Wholesale Club and a few health care stocks to deliver explosive returns, while least-crowded names in energy and real estate lagged due to demand shocks.
    • 2024: The AI infrastructure boom propelled heavily owned names such as AppLovin, Axon, Vertiv and Constellation Energy (aided by strategic power agreements) to standout performance. Notably, fundamentally strong crowded names like Horizon Therapeutics, Deckers and e.l.f. Beauty also contributed to most-crowded quintile strength in 2024, showcasing that crowding, when aligned with solid fundamentals and macro trends, can yield substantial upside.
  • Weak-market analysis clarifies the risk-management value of the signal. Conditioning long-short returns on Russell 3000 monthly returns shows the spread is largest in negative markets: when the index return is below 0%, the least-crowded minus most-crowded portfolio earns 1.5% per month, with a t-statistic of 2.9. This helps explain why crowding is most useful as a drawdown-control input rather than a standalone return forecast (Exhibit 4).
  • The cumulative return evidence is economically meaningful but should be interpreted with statistical caution (Exhibit 5). The CAGR spread between least crowded (Q1) and most crowded (Q5) groups remains close to 5% for equal-weighted quintiles and around 4% for value-weighted quintiles up to a 12-month holding period, even though monthly Q1–Q5 long-short returns are not statistically significant over the full sample.
  • Small-cap results point to a liquidity channel. The return differential peaks at roughly 6.1% value-weighted and 5.9% equal-weighted over a six-month holding period, consistent with less liquid stocks being more sensitive to institutional positioning and unwind risk (Exhibit 6).

 

Exhibit 3: Compound Annual Growth Rates by Crowding Quintile

Exhibit 3: Compound Annual Growth Rates by Crowding Quintile

Source: ClearBridge Investments January 2005 to June 2026. Note: Q1 2026 and Q2 2026 returns are annualized quarterly figures. Annualization can magnify short-term moves, so these observations are not directly comparable to full calendar-year returns. CAGRs shown for each crowding quintile (Q1 = least crowded, Q5 = most crowded). The left panel is ordered chronologically by year, while the right panel is ordered by the (Q1–Q5) CAGR (highest to lowest).

 

Exhibit 4: Performance of Least-Crowded vs. Most-Crowded Stocks by Russell 3000 Return Threshold

Exhibit 4: Performance of Least-Crowded vs. Most-Crowded Stocks by Russell 3000 Return Threshold

Source: ClearBridge Investments January 2005 to June 2026. Each threshold contains sufficient monthly observations for interpretation. Even the most restrictive weak-market cutoff, when the Russell 3000 return was below -6%, includes 25 months. The 0% and 2% thresholds are emphasized as weak-market cutoffs. The <0% threshold captures negative market months with 78 monthly observations, while the <2% threshold captures negative and low-return months with 121 monthly observations.

 

Exhibit 5: Cumulative Returns of Least-Crowded vs. Most-Crowded Stocks

Exhibit 5: Cumulative Returns of Least-Crowded vs. Most-Crowded Stocks

Source: ClearBridge Investments. January 2005 to June 2026. Shows both equal-weighted (EW) and value-weighted (VW) measurements. Although monthly Q1–Q5 long-short returns are not statistically significant over the full sample, the compounded CAGR spread remains meaningful.

 

Exhibit 6: Return Differential Between Least- and Most-Crowded Stocks

Exhibit 6: Return Differential Between Least- and Most-Crowded Stocks

Source: ClearBridge Investments. January 2005 to June 2026.

 

Hit Rate Analysis: Stock Level Evidence of Crowding Risk

To assess the predictive usefulness of crowding at the stock level, we evaluate the 12 month hit rate of the most-crowded quintile (Q5) relative to the Russell 3000. On average, 56.7% of stocks in Q5 underperformed the benchmark. Moreover, the dispersion of outcomes within Q5 was substantial: median underperformers lagged the benchmark by −29.2%, while median outperformers exceeded it by +24.8%. This asymmetry highlights that although crowded stocks can occasionally outperform, the typical downside among losers is large, reinforcing crowding’s role as a risk signal rather than a directional alpha factor.

We also regressed the monthly long-short returns of the Active Crowding factor against the Fama-French three-factor (FF3), five-factor (FF5) and six-factor (FF6) models, with the momentum control following Carhart (1997).2 Here we show the annualized regression alphas, factor exposures and t-stats (Exhibit 7).

The Active Crowding factor had a modest annualized alpha of 3.6% (a significant t-stat of 1.97) with respect to the FF6 model, and an annualized alpha of 5% (a significant t-stat of 2.37) with respect to the FF3 model, illustrating that the factor contained incremental risk information not fully captured by standard factor exposures. The Active Crowding factor had negative loading on the Momentum factor and positive loading on Value, Profitability and Investment factors. Thus, following stocks that are less crowded effectively tilted toward firms with high Value, high Profitability, low Investment and low prior returns.

Exhibit 7: Regression Coefficients of the Active Crowding Factor Against Fama-French Factor

Exhibit 7: Regression Coefficients of the Active Crowding Factor Against Fama-French Factor

Source: ClearBridge Investments. January 2005 to June 2026. Mkt-RF: Market Minus Risk-Free; SMB: Small Minus Big; HML: High Minus Low; RMW: Robust Minus Weak; CMA: Conservative Minus Aggressive; and MOM: Momentum.

What Explains Active Crowding?

Institutional constraints such as performance pressure and redemption risk often push long-horizon managers toward short-term behavior. Capital can exit swiftly when near-term performance lags, discouraging patient investing and reinforcing herding behavior. In equilibrium, herding increases demand for the stocks being crowded, pushing demand above the optimal level. This raises prices, lowers expected returns and at times makes such names vulnerable to severe drawdowns.

Conversely, some stocks tend to remain underinvested due to herding pressure — these stocks typically trade at lower prices, offer higher expected returns and attract less attention from career-concerned managers. Based on the dynamics of career-concern-driven herding, we expect these underinvested stocks to outperform crowded names over time.

Properly measuring active crowding therefore offers powerful insights into latent portfolio risk. Active crowding diagnostics can help managers avoid consensus traps, improve diversification and reduce downside vulnerability, especially when market liquidity deteriorates or crowded trades unwind.

Managing Crowding Risk

We believe neutralizing exposure to crowded trades is essential for effective risk management. Crowding is a frequently mispriced risk factor because it is difficult to observe directly. Crowded stocks tend to be consensus trades, marked by high expected returns and near-term stability. However, the buildup of forward-looking risk stemming from ownership dynamics (speculators versus long-term holders), sentiment imbalances (shared beliefs) and excessive optimism is often overlooked by conventional risk tools looking at trailing beta and volatility.

While the least-crowded stocks do not consistently outperform the benchmark, the most-crowded names consistently underperformed; this underperformance increased across longer holding periods up to 12 months and remained statistically meaningful (Exhibit 8).

Exhibit 8: Benchmark-Relative Returns of Most- and Least-Crowded Quintiles (U.S.)

Exhibit 8: Benchmark-Relative Returns of Most- and Least-Crowded Quintiles (U.S.)

Source: ClearBridge Investments. January 2005 to June 2026. Shows benchmark-relative returns versus the Russell 3000 Index of the most crowded (Q5) and least crowded (Q1) quintiles. Monthly data compounded to annual periods (1/31/2005–6/30/2026).

 

We next test whether the U.S. crowding patterns persist outside the domestic universe.

Factor Performance Results: Global

Expanding the analysis to the S&P Global BMI and FTSE Global Index universes, we observe broadly similar patterns to those seen in the U.S. case. As in the U.S., the most-crowded quintiles generally exhibit more negative return skew than the least-crowded quintiles, with the composite signal showing the strongest and most consistent separation. Most notably, the most-crowded quintile (Q5) underperformed the MSCI ACWI IMI benchmark by approximately –3.0% to –3.5% across one- to 36-month horizons, with value-weighted returns slightly worse, reaching –3.62% at the 12-month horizon (Exhibit 9). This underperformance increased up to 12 months and remained meaningful over longer horizons.

Exhibit 9: Benchmark-Relative Returns of Most- and Least-Crowded Quintiles (Global)

Exhibit 9: Benchmark-Relative Returns of Most- and Least-Crowded Quintiles (Global)

Source: ClearBridge Investments. January 2005 to June 2026. Shows benchmark-relative returns versus the MSCI IMI Index of the most crowded (Q5) and least crowded (Q1) quintiles. Monthly data compounded to annual periods (1/31/2005–6/30/2026[J).

Conclusion

The updated evidence shows that crowding is best understood as a forward-looking risk signal: it highlights where ownership, sentiment and momentum have combined to create downside vulnerability.

For portfolio managers, the practical takeaway is not to mechanically avoid crowded stocks, but to recognize when consensus positioning has become an uncompensated risk. Active crowding analysis can complement fundamental research by identifying positions where sentiment, ownership and momentum appear extended, while also highlighting less-crowded areas where return expectations may be less fully priced.

Used this way, crowding is not a timing tool or standalone alpha factor, but a portfolio construction input that can improve diversification, risk awareness and resilience through market cycles.

Appendix

Subfactor Construction Notes

One construction detail merits mention: for the Top 20 Active Manager Overweights subfactor, we observe a strong positive correlation (exceeding 0.7) between the raw number of overweights and company size. The analysis is conducted fund-by-fund using a global holdings FactSet dataset covering approximately $13 trillion in AUM across roughly 16,000 funds (2005–2026), each with at least $100 million in AUM; assets are concentrated in the U.S. and Europe, and nearly all funds update holdings within six months. Each fund is mapped to its benchmark using FactSet data; where no direct mapping exists, we assign the benchmark with the lowest active share across a predefined set of commonly used global indices. To correct for this size bias, we adjust the observed Top 20 Active Overweight Count using a bucketing regression approach. The Top 20 Active Overweight Count is regressed on log market capitalization within four size buckets — small, mid, large and mega-cap (the top 10% of large cap stocks) — drawn from our investable universe. The unadjusted count produces a −6.1% Q1–Q5 CAGR spread (Jan 2005 – June 2026), consistent with the raw measure being dominated by size exposure; after controlling for size the relationship reverses, with the low adjusted quintile outperforming the high adjusted quintile by 2.3%.

Related Perspectives

Deconstructing High-Quality Equity Outperformance
High-quality firms tend to outperform both low-quality firms and the broader market.
Ingredients for a Sustained Value Run
Farhan Mustafa, Sam Peters and Jeff Schulze debate what it will take for value stocks to maintain their current market leadership.
More
  • 1  Jain, K., & Jiao, D. (2025). Exploiting Myopia: The Returns to Long-Term Investing. SSRN
    2  Fama, Eugene F., and Kenneth R. French. 1993. “Common Risk Factors in the Returns on Stocks and Bonds.” Journal of Financial Economics 33 (1): 3–56.
    Fama, Eugene F., and Kenneth R. French. 2015. “A Five-Factor Asset Pricing Model.” Journal of Financial Economics 116 (1): 1–22.
    Carhart, Mark M. 1997. “On Persistence in Mutual Fund Performance.” Journal of Finance 52 (1): 57–82.

  • Past performance is no guarantee of future results. Copyright © 2026 ClearBridge Investments. All opinions and data included in this commentary are as of the publication date and are subject to change. The opinions and views expressed herein are of the author and may differ from other portfolio managers or the firm as a whole, and are not intended to be a forecast of future events, a guarantee of future results or investment advice. This information should not be used as the sole basis to make any investment decision. The statistics have been obtained from sources believed to be reliable, but the accuracy and completeness of this information cannot be guaranteed. Neither ClearBridge Investments, LLC  nor its information providers are responsible for any damages or losses arising from any use of this information.

more