Ex‑Citadel PM on scaling multi‑manager alpha
A social post from ex‑Citadel PM Brett Caughran argued that multi‑manager funds have found ways to scale alpha by exploiting market inefficiencies, countering assumptions that AI alone will flatten trading edges. The thread was widely shared and discussed among trading practitioners as a competitive-intelligence signal. (x.com)
A former Citadel and D. E. Shaw portfolio manager says multi-manager hedge funds can still scale stock-picking edge by widening coverage, not just by adding artificial intelligence. (threadreaderapp.com) Brett Caughran posted a 25-part thread on September 26 laying out a hiring plan he drafted in summer 2020 while interviewing with a large multi-manager fund. He wrote that his market-neutral healthcare process capped out at about 40 stocks for one investor, so he proposed a five-person team to expand coverage and increase “idea velocity.” (threadreaderapp.com) In that plan, Caughran said the United States healthcare universe had 221 stocks trading more than $5 million a day with market capitalizations above $100 million. He said three senior idea generators could manage a portfolio above $1 billion if more than 75 percent of risk came from company-specific factors rather than broad market moves. (threadreaderapp.com) Multi-manager funds run many small teams, often called pods, inside one firm. Goldman Sachs said in a 2024 note that the sector’s headcount rose from about 5,300 in 2015 to more than 18,000 in 2023 as these firms expanded that pod model. (goldmansachs.com) Caughran’s argument was that scale comes from specialization: one analyst follows one slice of an industry closely enough to trade quickly when prices move away from fundamentals. He wrote that a coverage model works better than a generalist model in factor-constrained, market-neutral investing because teams need to be ready to trade dozens of names at any time. (threadreaderapp.com) That view cuts against a simpler version of the artificial-intelligence trade, which holds that better models will erase informational advantages across public markets. Caughran instead described a workflow in which humans narrow the field to a few key business drivers and compare them with the expectations already embedded in a stock price. (threadreaderapp.com) Caughran is now founder of Fundamental Edge, a training firm for buy-side analysts, and says he retired from managing money in 2021 after 13 years at Maverick, D. E. Shaw, Citadel, Bell Rock, Two Sigma, and Schonfeld. Institutional Investor reported in March 2023 that he had previously run a $1.5 billion book with a nine-person team. (fundamentedge.com) (institutionalinvestor.com) The backdrop is an industry that has kept getting bigger even as fees, data costs, and computing budgets have risen. Public filings tracked by WhaleWisdom show Millennium reported about $571.1 billion in discretionary assets on a February 3, 2026 Form ADV, while Point72 reported about $220.9 billion on a January 30, 2026 filing. (whalewisdom.com 1) (whalewisdom.com 2) Other practitioners make a narrower point: artificial intelligence may help summarize documents, process data, or speed up research, but it does not remove the need for position sizing, risk limits, and sector judgment inside a pod. Point72 says on its website that it deploys discretionary long-short equity, systematic, and macro strategies under one platform, a structure built around many distinct decision-makers rather than one central model. (point72.com) The reason the post traveled is that it treated hedge fund scale as an organizational problem, not a software problem. In Caughran’s telling, the edge comes from putting more trained eyes on more mispricings, then wiring those analysts into a risk-controlled machine that can act fast. (threadreaderapp.com)
