Herding behaviour in investing: how following the crowd amplifies market cycles

Markets aggregate information from many independent participants. When participants stop forming independent judgements and begin imitating each other instead, that aggregation breaks down. Devenow & Welch (1996) documented this phenomenon—herding behaviour—and its consequences: amplified market cycles, inflated valuations, and sharp corrections. For individual investors, following the crowd feels safe. The evidence suggests it is often the opposite.

What herding behaviour is

Devenow & Welch (1996), in Rational Herding in Financial Economics, distinguish two forms of herding. Rational herding occurs when investors free-ride on others' information: if many informed participants are buying an asset, it is rational to infer they know something worth knowing, and to follow. Irrational herding occurs when investors conform simply to avoid the regret of deviating from the crowd—not because they believe the crowd is informed, but because the psychological cost of standing apart feels higher than the risk of being wrong together.

Robert Shiller's speculative feedback loop mechanism describes how herding can sustain itself: rising prices attract attention, attention attracts buyers, buyers push prices higher, which attracts more attention. The loop continues until valuations disconnect from fundamentals far enough that the correction becomes inevitable.

How it manifests in investing

Herding appears in investment behaviour in several observable patterns.

Chasing recent fund performance. Capital flows disproportionately into funds that performed best in the recent past. This concentrates money in funds that are already expensive, pushing their prices higher and creating conditions for mean reversion. The investors who arrived last absorb the correction.

Selling during market panics. When markets fall sharply, each investor's decision to sell validates others' decision to sell. The exit becomes self-reinforcing. Investors who sell during panics often do so near the trough—crystallising losses at precisely the point where expected future returns are highest.

Asset class bubbles. Herding concentrates capital in fashionable assets—technology stocks in 1999, residential property in 2006, speculative assets in more recent cycles. The concentration amplifies valuations beyond what fundamentals support. The correction, when it comes, is proportional to the overvaluation that herding created.

The cost

Frazzini & Lamont (2008), Dumb Money: Mutual Fund Flows and the Cross-Section of Stock Returns, Journal of Financial Economics, examined the relationship between retail fund flows and subsequent returns. They found that stocks attracting the highest retail inflows underperformed by approximately 3 percentage points per year on average—a direct measure of the cost of herding into recent winners. Investors who bought at cycle peaks, driven by the momentum of others' buying, absorbed the correction that followed. Bubbles inflated by herding do not deflate gradually; they correct sharply, and the investors most exposed are those who arrived latest.

What helps

Systematic strategies execute based on defined signals—quantitative measures of momentum, valuation, or quality—rather than social proof or the observable behaviour of other market participants. The strategy's rules have no mechanism for detecting what the crowd is doing and no instruction to follow it. This does not mean systematic strategies are immune to all market dislocations—no strategy is—but they do not amplify cycles by definition-following behaviour in the way that sentiment-driven discretionary investing can.

Herding behaviour in pfolio

pfolio's systematic strategies execute based on defined signals—quantitative measures of momentum and volatility applied to each asset in the universe—rather than on social proof, fund flow data, or the observable behaviour of other market participants. The strategy's rules have no mechanism for detecting what the crowd is doing or for following it. This does not make systematic strategies immune to all market dislocations, but it does mean that allocation decisions are made on the basis of objective price information, not on the basis of what is popular. For a full description of the signal methodology, see how we build portfolios.

Related articles

• Overconfidence bias
• Confirmation bias
• Recency bias