Extension Strategies: Managing More Complex Risk to Uncover Greater Reward

How Extension Strategies Unlock Potential

Our proprietary quantitative stock selection models evaluate thousands of stocks per day. For any given investment universe, we split stocks into quintiles and rank them in terms of anticipated performance—and in general, these ranking models have been as effective at identifying underperformers as they have outperformers. For example, in a global developed markets quintile spread covering the past 20 years—where the first quintile represents the highest anticipated performance and the fifth represents the lowest—our highest-ranked stocks outperformed the equal weighted market by 4.96%, while the lowest-ranked stocks underperformed by 6.48% (Exhibit 1).

However, in long-only portfolios, our ability to express the full breadth of these views is limited. We can express our positive views by holding the highest-ranked stocks in a portfolio—but often our only way to express negative views is to simply not hold the lowest-ranked stocks, since the amount of underweighting allowed is determined by the benchmark weight of a stock. Currently, the median benchmark weight of all stocks in the MSCI World Index is just 0.03%, which leaves very little potential for reward when our negative views prove correct—that is, when the lowest-ranked stocks we do not own go on to underperform the wider market.

Extension strategies relax this limitation by allowing a certain percentage of capital to be used for short-selling stocks on the lowest end of our ranking spectrum. The proceeds from short-selling can then be invested into the highest-ranking stocks, offering additional long-only exposure: for example, 130% long and 30% short exposure in a 130/30 strategy, or 50% short and 150% long exposure in a 150/50 strategy. This is a particularly appealing feature for long-term investors who may find themselves running out of capital to express positive views.

In effect, long-only portfolios are augmented by the addition of a market-neutral, long-short portfolio that represents a certain percentage of the total capital invested. With additional capital invested, active exposure increases—along with the potential for higher returns. These portfolios are not constructed separately, but together, promoting an optimal consideration of return, liquidity, and risk constraints. The same amount of asset-owner capital is invested and the net market exposure remains constant, with a beta of around 1—but implicit leverage is employed, which can magnify both potential returns and potential losses.

Why Now?

This differentiated approach to portfolio construction fell out of favor in the early 2000s due to a range of factors, including the quant quake of 2007; record-high market volatility stemming from the global financial crisis; increased regulation; and fees that some investors found difficult to justify. But today, we are seeing renewed interest in extension strategies due to their compelling track record of long-term performance and their ability to address the unique challenges of investing in a highly concentrated market.

Today’s active managers are compelled to devote substantial client capital to the Magnificent Seven stocks that make up over 22% of global markets.³ The risk of being structurally underweight in these market leaders is unacceptably high, as not holding these stocks would result in significant underperformance to most benchmarks. As a result, active capital is being pulled toward these stocks and drawn away from compelling opportunities identified elsewhere in global markets. 

Additionally, the sheer size of today’s market leaders has reduced overall underweighting potential. In a long-only developed markets portfolio, underweights of 1.5% or more versus the MSCI World Index represent only nine stocks—all of which are US large caps. In this environment, lifting the long-only constraint means substantially expanding managers’ ability to express high-conviction positions in stocks they expect to underperform from nine stocks to thousands (Exhibit 2).

Put simply, extension strategies offer a way to increase risk, or tracking error, by leveraging what quants are good at: stock selection. By applying our insights from across the ranking spectrum, we can build portfolios that mirror the full scope of our convictions. The only other way to increase tracking error without extension strategies is to take macro bets against the make-up of a benchmark—for example, by being overweight or underweight in terms of country, size, or market cap. Such bets offer less breadth (fewer choices) and, as such, may offer fewer opportunities to diversify sources of potential alpha. For active managers of all stripes, these bets can also introduce significantly higher volatility and unpredictability to active return patterns, which can lead to lower information ratios (returns per unit of risk taken)—an idea we explore more deeply in Tracking Error: Myth vs. Reality. 

The Risks Inherent in Shorting

Identifying underperforming stocks is one thing—but short-selling stocks is a riskier endeavor than underweighting or eliminating them from a portfolio. This is due to the inherent unpredictability of our lowest-ranking stocks and the theoretically unlimited upside risk—or unlimited loss—that comes with short-selling.

In general, what the lowest-ranked stocks have in common is heightened unpredictability. These stocks are typically further down the capitalization spectrum (since negative views on large stocks are often sufficiently expressed by not holding them in a portfolio), and these smaller businesses tend to exhibit more unpredictability than larger businesses do. Their futures are generally more uncertain (resulting in elevated volatility); they tend to have weaker or deteriorating financials (resulting in weak momentum); and they tend to be more vulnerable to sentiment changes that can cause sudden, significant upswings in performance.

Finally, because these weaker-performing stocks are often already heavily shorted, they are prone to sudden surges in price, or short squeezes, whereby a rush of short-sellers buy shares to close out their positions in anticipation of a stock’s price rising further. Short squeezes are relatively rare events—but the historic GameStop short squeeze of 2021 drew attention to how quickly they can happen in the information age and how detrimental they can be to short-sellers.⁴

To illustrate the impact of these hidden but elevated risks, we ranked one set of stocks in two different ways: from high to low 12-months-prior momentum and from high to low volatility, where the lowest-momentum and the highest-volatility stocks represent those that would typically be most suitable for shorting (Exhibit 3). We then evaluated each quintile’s average active returns on the best 1% of days (of 9,412 trading days) compared to its average annual returns from 1989–2025. This breakdown reveals wide gaps between the typical performance of these lowest-ranked stocks and their performance on the strongest days.

Low-momentum stocks in the fifth quintile—representing the 20% of stocks with the weakest prior 12-month total return—did tend to underperform by 3.18% on average over this 36-year period. However, their performance in the top 1% of days averaged 2.71%, far outstripping the performance of the other quintiles. On their single best day—9 November 2020, coinciding with the announcement of a COVID-19 vaccine⁵—the fifth quintile’s outperformance versus the market reached 7.65%.

Similarly, over the same 36-year period, highly volatile stocks in the first quintile saw negative average active returns of 1.44% over the long term—but their performance on the best 1% of days reached 3.18%, far outstripping that of the other quintiles. On their best day—3 January 2001, when the Federal Reserve announced a surprise 50-basis-point rate cut aimed at combating slow economic growth⁶—these stocks outperformed the market by 6% on average, a figure far higher than the other quintiles.

Over the long run, while highly volatile stocks or stocks with very weak prior returns do exhibit unattractive characteristics that make them suitable for short-selling, they can also present elevated upside risk at unexpected times.

A Balanced Approach Is Key

Because of this unpredictability, we apply a number of additional portfolio construction risk mitigation measures to extension strategies beyond those used for long-only strategies (Exhibit 4). The goal is to build a short basket with exposures across an array of capitalization, industry, and macro drivers, while factoring in thorough assessments of unattractive stock characteristics across multiple stock selection dimensions—for example, our views on growth outlooks, relative valuation properties, and in-depth textual news sentiment analysis. Furthermore, we recognize that position sizing takes on greater significance on the short side and for that reason we calibrate again using a number of key additional measures. Here we aim to mitigate over-exposure to stocks with high levels of idiosyncratic risk, using crowding indicators to further identify stocks that could be most vulnerable to short squeezes.

Another way to limit shorting risk is to utilize a 130/30 construct, which caps shorting at a modest 30%. Additional shorting (e.g., 50% in a 150/50 construct) can be more lucrative, offering greater exposure to stock selection insights and higher tracking error—but it also introduces more positions, less transparency, and more complexity to the management of the short side of the portfolio. With most ‘short’ stocks smaller in size, incremental capital is drawn to a greater number of smaller companies, pushing the limits of the risk mitigation enhancements previously outlined. In our view, the 30% is the sweet spot for investors seeking to increase active return potential and tracking error while still limiting their exposure to shorting risk.

Conclusion

We expect extension strategies to see significantly more interest in the years ahead. But we do not believe all shorting approaches are created equal. In our view, a balanced approach is necessary to help ensure that portfolios are not exposed to undue risk—particularly in an environment where macroeconomic and geopolitical uncertainty can make changes in investor sentiment (and stock price behavior) more unpredictable and extreme.