This article creates a novel framework to explore a number of alternative risk premia (ARP) strategies to assess the impact of crowding on subsequent strategy performance. The aim is to provide investors with a better understanding of the consequences of crowding in order for them to design better strategies and improve portfolio construction and risk management.

A cornerstone of the author’s approach is the concept of divergence and convergence premia. Divergence risk premia feed on themselves, systematically increasing returns to winning strategies and imposing losses on losing strategies. A prime example is momentum, where investors buy into outperforming assets and sell underperforming ones. Strong flows into momentum strategies create a positive feedback loop, pushing up (down) expensive (cheap) prices and positively fueling the outperformance. The more crowded the trade, the more prices diverge from fundamentals. Because momentum strategies are not based on fundamentals, investors have no way of knowing when a trend has run its course and may suffer the consequences of a sudden and nasty correction.

Convergence premia, however, systematically reduce the spread between undervalued and overvalued assets. A value strategy, for example, buys undervalued assets and sells overvalued assets. Significant flows into such a strategy reduce the valuation spread. In convergent strategies, spreads narrow to the point where further profits are small or nonexistent, and the opportunity disappears.

To test and quantify these ideas, the author analyses a range of risk premia strategies, including value (book to price), size (market cap), momentum (over 12 months), quality (return on assets), and low beta. To do so, he chooses liquid assets with long price histories: equities, commodities, and currencies.

Portfolios are constructed by taking a long position in top-ranked assets and a short position in bottom-ranked assets. The author uses excess co-movement across top and bottom assets to indicate crowding.

Divergence premia within equity, commodity, and currency markets underperform after crowded periods.

An equity momentum strategy, for instance, has positive returns over the first month after both crowded and uncrowded periods. Returns are very similar under either condition.

When crowding is very high, however, the strategy underperforms, losing 2.25% in the first month alone in the author’s sample. However, momentum strongly outperforms after low levels of crowding, returning 4.6% after 6 months and 6.9% after 12 months.

These findings are replicated across all divergence premia, supporting the theory that divergence strategies cause a coordination problem for investors. That is, investors cannot tell in real time how many other investors are invested in a strategy and are thus at risk of negatively affecting each other’s performance.

Convergence premia, however, behave in the opposite way. Lower levels of crowding are linked to poor future performance, and high performance appears to follow high crowding.

The author’s results undermine the assumption that the crowding of factor strategies always leads to negative performance. Using the concepts of divergence and convergence, the author shows that different types of strategies actually respond differently to investor flows.

This finding has implications for risk management and multifactor portfolio construction that could be valuable to investors.

In particular, investors should consider how best to manage the risk in strategies with divergence dynamics. Volatility targeting is one way to improve risk-adjusted returns of such strategies as momentum and betting against beta. However, returns are probably not improved by managing the volatility of convergence strategies because volatility does not spike when the trades become crowded.

Lastly, the author suggests that the predictive nature of his crowding framework for ARP performance could serve as a starting point for investors seeking to time factor-based investments.