The panel, moderated by John Delano, Managing Director and Head of Research and Analytics at Commonfund, featured Jens Foehrenbach, President and Chief Investment Officer of Graham Capital Management, and Mark Refermat, who leads the machine learning portfolio strategy team at Voleon Capital Management.
With equity valuations stretched and bond market uncertainty lingering, the panel framed its discussion around a clear premise: in today’s environment, strategies that generate returns uncorrelated to equities, fixed income, and illiquid alternatives deserve renewed attention. But uncorrelated to markets is not the same as uncorrelated to each other – and building a truly diversifying portfolio requires both. The pairing of a global macro manager and a quantitative equity manager illustrated this well: while both can generate returns independent of market direction, their approaches differ significantly in the instruments they trade and the logic behind their positions, making them distinct and complementary sources of diversification.
Foehrenbach opened by reflecting on how the macro landscape has shifted since the era of quantitative easing. The end of that period has created a richer opportunity set defined by policy divergence, regional economic fragmentation, and a reassessment of risk premia across currencies, rates, and emerging markets. He described the current moment as a transition phase – while last year’s dislocations created fertile ground for macro strategies, early 2026 has brought a challenging backdrop in which signals are less directional and opportunities require greater selectivity.
Refermat approached the same environment from a different angle. Voleon’s strategy is designed to remain market‑neutral, capturing returns from dispersion – the performance gap among individual companies – rather than from broad, macro-level market moves. He noted that dispersion has been elevated partly because macro forces are affecting firms unevenly, depending on their supply chains, cost structures, and access to financing. For active traders who generate alpha across individual stocks, the current U.S. equity environment has been especially supportive. In his view, this combination of heightened dispersion and robust liquidity creates a compelling backdrop for systematic, market‑neutral approaches that thrive on idiosyncratic signals rather than macro directionality.
A central theme emerged regarding what machine learning can bring to investing for allocators searching for uncorrelated return streams. Traditional quant strategies rely on human-generated hypotheses that are then tested against data. This approach is transparent but can limit alpha generation to what a researcher can imagine. In a market where many quantitative signals have become crowded or commoditized, that constraint narrows the opportunity set.
Machine learning expands that frontier by uncovering often-complex, nonlinear relationships across datasets, often targeting subtle effects that a researcher starting with a preconception of an idea would be unlikely to search for or test. Refermat illustrated this with a simple example: satellite imagery of parking lots. A traditional quant might extract two signals from that data – which retailers have the most cars, and who is trending up. Machine learning expands the frontier of what can be found by searching a much larger and more complex space of potential relationships than a human researcher would typically consider. A traditional quant often begins with a specific economic intuition and tests it. A machine‑learning model begins with the data and explores a far wider set of interactions. By analyzing more data simultaneously, finer distinctions emerge – and with them, a source of nuance that can drive returns.
The panelists were candid about the difficulty of this: failure rates are typically high, and the process of refining a strategy can be slow and nonlinear. For allocators seeking truly uncorrelated returns, that difficulty is part of what makes such alpha opportunities durable, and barriers to entry help preserve the opportunity set. A genuine edge can take years to build and refine, and more alpha can accrue to firms with deep, hard-won expertise.
Risk management also emerged as a priority, with both panelists underscoring the need for explicit frameworks and the discipline to enforce them. They pointed to practices such as risk limits for portfolio managers, daily risk‑committee oversight, and automated monitoring to maintain market, sector, and factor neutrality. These structures are not just operational guardrails; they are essential to ensuring that each strategy behaves as intended and delivers the uncorrelated return profile investors expect.
Both managers acknowledged that attracting and retaining top talent is a persistent challenge, but that culture can be a meaningful differentiator. Voleon recruits heavily from the academic community, prioritizing PhD researchers trained to tackle unsolved problems and rigorously test their ideas. By mirroring academia’s collaborative, peer‑review ethos—where ideas are strengthened through challenge rather than protected from it – the firm appeals to researchers seeking that level of intellectual rigor.
Graham Capital takes a different but equally intentional approach. With 25 discretionary teams, the firm encourages ongoing dialogue across groups—unusual in an industry where individual portfolio managers often guard their positions closely. A culture of open exchange strengthens the firm’s overall investment process and supports talent retention by creating a collaborative learning environment.
In a market environment where traditional return sources look expensive and uncertainty around forward return is especially high, the panel offered a compelling reminder that alpha still exists, although it may be harder to find and to sustain. For endowment and foundation investors, the lesson about the value of genuine diversification was clear: understanding managers’ individual strategies and why their return drivers are different, how they might behave under stress, and the culture and process behind the numbers.