Created
Dec 17, 2024 11:05 PM
Tags
Notes
A new model has overtaken the hedge fund industry: the multi-strategy “pod shop” that features dozens of siloed manager teams. The financial news has been full of stories about both their fundraising success (Millennium recently raised another $10 billion) and their impressive investment returns (Citadel’s flagship multi-strategy fund has delivered annual returns of 19% since 1990).
These pod shops share a few distinctive features, as Giuseppe Paleologo, a former risk director at Millennium and Citadel, explained recently on Bloomberg’s Odd Lots Podcast. They manage a large number of teams (Millennium reportedly operates close to 300). Money is constantly being reallocated among the pods on a seemingly simple basis: alpha generation. “If they perform well, to give them more capital,” says Paleologo. “If they don't perform well, to take capital away from them or let them go.” The individual pods run with very little market and factor exposure, and remaining undesirable exposure is usually hedged out by a centralized risk team. These funds are run at very high leverage levels, amplifying the uncorrelated alpha return streams generated by individual PMs.
With these facts in mind, we had a simple idea. What if we took all the actively managed mutual funds and ETFs and treated each as its own pod, using our own internal risk model to separate the funds’ alpha from their systematic market and factor exposures? Could we build a sort of poor man’s pod shop—or at the very least use the exercise to better understand the model? We don’t think there’s any reason to think that Fidelity or Capital Group’s best managers aren’t equally skillful to Citadel’s or Millennium’s.
We know from decades of research from SPIVA that most active managers struggle to beat the market and that outperformance of the index is not persistent over the long run. Making pod shop lemonade out of the lemons of the world of active management must, therefore, rely on either shorter time horizons (perhaps active managers have hot streaks of a few months that we can bet on) or the risk model they use to manage and combine the streams is enormously powerful.
Short-Term Persistence of Alpha
We can test the first possibility empirically. We analyzed alpha persistence across 3,182 actively managed mutual funds and ETFs from 1997 to 2024. Using rolling time-series regressions of fund returns against 13 factor return streams, we estimated fund-level factor exposures (betas). These factor betas included market, value, size, momentum, and several others representing some of the most well-studied academic risk factors.
After estimating these exposures, we measured alpha, which represents a fund’s incremental performance unexplained by factor betas. As illustrated in Figure 1, alpha over any investment period was calculated as the total return (net of fees), minus the risk-free rate (Fed funds) and factor-implied performance. Factor-implied performance was derived by applying each fund’s beta coefficients to the actual realizations of each respective factor.
Figure 1: Illustration of Fund-Level Performance Attribution |
