Generic Exposures: Not All Gold Glitters!

Value and Quality Never Go Out of Style

Active managers aim to deliver excess risk-adjusted returns by identifying mispriced securities, constructing portfolios that maximize exposure to these insights, and carefully managing risks and transaction costs. Each manager will look for companies that are cheap, high quality, display strong fundamental growth prospects, and may benefit from favorable market dynamics - although the balance placed on these focuses will depend on a given manager’s style, skill in each respective area, and investment horizon.

Years of empirical research have identified several company characteristics and behavioral anomalies that lead to predictability of excess returns along those four key dimensions. These are abundantly documented and analyzed in journal publications and have served as the foundational pillars for many active managers’ investment processes. While some managers simply offer a cost-effective way of obtaining exposure to these well-known thematic drivers, the overwhelming majority stress a bevy of their own proprietary flavors. Their insights are still rooted in the same general themes but offer superior outcomes by better applying existing data or by tapping into newer, timelier, or more granular information about company fundamentals and actions of market participants.

Getting the Right Answers Starts with Asking the Right Questions

With increasing volume and variety of new “alternative” data sets and the growing sophistication of modeling approaches, active managers are expected to derive more unique and novel alpha insights. It would seem natural to then conclude that their forecasts and portfolios would, correspondingly, show less exposure to well-known constructions that are now deemed generic. Some would even go as far as to say that these exposures should be as low as zero and anything above implies a less sophisticated investment process, one riding on the coattails of generic premia.

Needless to say, anyone looking to invest in an active strategy would want to ensure that it offers unique alpha drivers. Traditional analyses (such as regressing a manager’s ex-post returns on those of generic factors, conducting cross-sectional regressions of forecasts on generic characteristics, and attributing portfolio risk to generic exposures) are at best incomplete and at worst misleading. Instead of focusing solely on measuring a manager’s exposure to generics, the questions and analyses should concentrate on whether these exposures reflect skill.

Don’t Judge a Book by Its Cover

An active portfolio may exhibit exposures to generic factors for several reasons, but only one of them is indicative of the “low proprietary skill manager” problem that is often reflexively assumed to be the cause. These reasons fall into two broad categories. 1) Generic exposures might be embedded upstream in the alpha and then flow through to the portfolio active weights. 2) Even with an alpha forecast entirely devoid of any generic exposures, the portfolio may still acquire such exposures downstream. Let’s consider these in turn.

Will the Real Alpha Signals Please Stand Up?

We begin with the most problematic of all possible scenarios. A manager’s alpha model could rely heavily on persistent linear bets on well-known asset pricing factors as the primary drivers of expected benchmark-relative excess returns. Such exposures would likely be substantial and stable over time, while the resulting return stream would exhibit significant volatility. This would indeed suggest that the manager has low proprietary alpha content. Aside from the aforementioned high and stable loadings on generics, some red flags associated with such managers include low investment in proprietary research efforts (such as human capital, data, technology stack, and an innovative research agenda) and an unwillingness to provide clear and coherent examples of proprietary components in their investment process, often under the guise of intellectual property protection (any “secret sauce” usually can still be intuitively explained in a way that won’t divulge IP). 

Alternatively, an active manager’s exposure to generics could result from explicit or implicit timing components that dynamically shift allocations to align with expected payoffs, an activity which requires considerable skill. Because generics are well known and subject to significant arbitrage activity, they have evolved to behave like risk factors — exhibiting high payoff volatility and carrying premia as compensation for the associated risks. Consequently, predicting changes in their payoffs presents potential alpha opportunities. For example, a manager might employ a proprietary method to model “risk-on” versus “risk-off” environments and then adjust generic exposures to align the portfolio with the expected flow of the “river of risk.”

Lastly, and perhaps most importantly, exposures to generics can result from interaction effects between proprietary alpha components and well-known factors. These interactions can arise in a linear model that includes interaction terms or within a sophisticated nonlinear model that has generic factors among its features. In other words, beyond just when (dynamically timing exposures), the manager can also identify the where and how — specific conditions under which these generics deliver higher and/ or more consistent payoffs. This added layer of insight effectively transforms a common factor into a powerful source of alpha.

Every Rose Has Its Thorns

Even if a manager’s alpha forecast is entirely cleansed of all well-known asset pricing factors, their portfolios may still acquire exposures to such generics downstream. This can happen for several reasons. Portfolio construction constraints often interact with alpha and benchmark properties. For example, a long-only portfolio may struggle to meaningfully underweight small-cap stocks and thus end up overweighting small caps, funded by underweights in large caps, inadvertently creating a bet on the size factor. Similarly, generic factors may be related to alpha dispersion. Whether by explicit design or as a natural outcome of forecast formation, alpha forecasts might be larger in magnitude in the tails of certain generic factors (e.g., more dispersion in less liquid securities). Coupled with other constraints, this can induce generic factor exposures, even if the alpha wasn’t linearly related to these factors.

As another example of generic portfolio exposure not driven by the alpha, consider a portfolio that has enjoyed significant benchmark-relative performance and low turnover due to stable alpha forecasts. Even if this portfolio’s alpha isn’t explicitly betting on generic momentum, it will gradually acquire such exposure unless the risk model or constraints prevent it.

Separating the Wheat from the Chaff

So how can asset owners better identify skilled managers in the context of generic exposures? Let’s start by considering the types of analyses best avoided, which also happen to be those that we see most frequently undertaken: 1) regressing fund performance on the returns of simple factors and 2) assessing the fraction of active risk attributed to generic styles.

While both approaches are easy to implement and may seem intuitive, they can result in misleading conclusions. Looking at the first approach, time series performance attribution may not provide an accurate assessment of a manager’s alpha (as measured by the intercept term) if the regressors are not investable. Care should be taken to ensure that they reflect long investment opportunity in a universe of sufficient liquidity. Additionally, the covariance of ex post payoffs does not necessarily indicate the presence of ex ante exposures. Consider the example of volatility exposure. Large exogenous shocks that disrupt the markets typically degrade the performance of active managers, whether systematic or discretionary. This is intuitive. Performing brain surgery during an earthquake is unlikely to yield good outcomes even for the best of surgeons. Because realized returns often suffer during periods of market turbulence and heightened volatility, a time series regression might suggest that the manager was “selling volatility,” even if their investment process did not involve such exposure ex ante. 

Looking at the second approach, decomposing active variance provides only a partial view of the drivers of portfolio positioning. Active variance contribution depends on the degree of exposure to a source of risk, its variance, and, to a lesser extent, the correlation of that risk with other drivers of co-movement. Because generic factors are inherently more volatile, they tend to dominate in risk decomposition analyses, even at modest active exposures. Proprietary components are lower volatility in return contribution and less exposed to broad negative events (i.e., exogenous shocks). A proprietary alpha signal that adds a steady excess return will make little contribution to a decomposition of active variance, even if it is a key driver of active weights. 

We recommend a more informative approach where the manager regresses their alpha on generic factors at each point in time and then separately tracks performance of 1) the portion of the alpha explained (“spanned”) by generics and 2) the residual. By construction, the unexplained residual relates to alpha not linearly related to any generic exposures. This residual component should show high efficacy, low risk and a high Sharpe Ratio. The fitted part would still potentially embed manager skill related to temporal variation in exposures to generic factors. Relative to the unexplained residual, it is expected to exhibit lower efficacy and higher volatility. Insofar as the spanned component could involve dynamic exposures to generic factors, assessment of the timing skill that it reflects requires benchmarking its performance relative to some static blend of generic factors. One option would be to use the ex post average of exposures from the regressions, but that implies ex ante foresight as to what those ex post exposures turned out to be. A fairer benchmark would be derived from some simple and transparent static weighting of generic exposures, e.g., a combination that would have generated equal ex ante estimated risk contributions.

Equipped with this separation of forecasts into residual and generic-spanned parts, the manager can assess the sensitivity of portfolio active weights to the two sources of return. Whereas a decomposition of active variance may be dominated by generic exposures, the active tilts of the portfolio (i.e., the portfolio active alpha shares) might more closely reflect the residual forecasts.

Sequentially peeling the layers as described above would allow the manager to provide clear evidence of the value that generic exposures bring to their portfolios and to show the extent to which they are driven by proprietary alpha insights.

Conclusion

Sophisticated systematic investors and consultants are naturally drawn to the narrative that “proprietary” signals offer superior risk-adjusted returns and are the exclusive domain of leading quant firms. Generic exposures within active portfolios are often viewed with suspicion, raising concerns about a lack of proprietary skill. However, as this paper has explored, their presence does not necessarily indicate a simplistic investment process. Instead, these exposures can often be a necessary byproduct of sophisticated alpha models and the interplay between alpha and portfolio construction. When these exposures are driven by proprietary insights, they can enhance excess returns, transforming generic factors into powerful alpha drivers. Stripping them away, while compelling from a marketing perspective, will discard valuable components of the investment process and lead to diminished returns and lost opportunities.

Understanding the true nature of generic exposures requires moving beyond analyses that focus solely on factor loadings or simplistic performance attributions. By relying on more refined but still intuitive methods, asset owners and consultants can better distinguish between managers who simply track generic factors and those who actively leverage them to generate true alpha.