Interview Summary: AI, Risk Management, and Building a Multi-Manager Fund

Performance and Investment Approach

• The guest reports 50% annualized returns with a 2.1 Sharpe ratio, while emphasizing that past performance may include luck and is not a promised future rate.
• He runs a highly leveraged portfolio—around 400% gross exposure—and therefore prioritizes downside risk, particularly on long positions.
• He makes hundreds of independent bets each year and says frequent investing “reps” help him recognize what drives a stock.
• His general approach is to focus on what works, make decisions quickly, and avoid becoming attached to an idea. For experienced investors, he suggests deciding within an hour or two—or moving on if the idea remains unclear.

How AI Supports the Investment Process

• AI translates and summarizes filings and news, then ranks items by importance, allowing him to focus on the most relevant handful rather than review a large volume of low-value information.
• A custom AI-powered risk system groups holdings into evolving baskets or “factors,” including categories the system identifies itself.
• The system updates correlations daily. Because relationships between stocks can change, a portfolio’s effective exposure to a basket can grow or shrink even when its holdings do not change.
• The tools have substantially changed his company-by-company workflow, reducing routine work and freeing attention for higher-value decisions.

Managing Risk and Portfolio Correlations

• The fund tracks both traditional factors and custom groupings, as well as correlations among portfolio managers.
• The guest checks manager-level correlations periodically and reviews unusual losses and individual portfolio risks.
• He says the core managers’ cross-correlation is about 0.15; estimates for a broader set of managers remain low but are based on a limited sample.
• He deliberately limits discussion of investment ideas among managers to help preserve independent strategies and low correlation.

Recruiting and Scaling Portfolio Managers

• The fund looks for verified live track records, not credentials. Managers have come from varied backgrounds, including programming, government, poker, finance, and other fields.
• It screens for fabricated or cherry-picked results and tests for hidden tail risks, such as strategies that rely on selling volatility or doubling down on losses.
• New managers start with small allocations. Capital increases depend on the number of bets, demonstrated risk control, performance, and correlation with the rest of the fund.
• Longer-term strategies may need six to nine months before an initial increase; faster-trading strategies can produce evidence sooner.
• The guest says managers are generally compensated through carry tied to profits above the risk-free rate, with terms tailored to the strategy.

Why Managers Might Join

• The fund offers backing, shared technology and staff, and lets managers avoid building their own operational infrastructure.
• The guest also sees potential tax-deferral benefits within a leveraged long-short fund. He describes using losses on some positions to offset gains elsewhere, while noting that tax outcomes depend on how positions and fund structures work.
• He argues that managers can retain a place in the fund’s investment “engine” even if their own strategy later loses its edge, though a portfolio manager could still be removed from managing capital.
• The fund reportedly manages about $480 million, has additional managers in trial, and has a waitlist for investors.

Investment Lessons and Habits

• He favors focusing on the key variable that drives a stock rather than doing exhaustive research for its own sake.
• For shorts, he has learned not to cling to a thesis when price action persistently contradicts it without a clear catalyst.
• He is open to buying stocks making new highs when the underlying case supports continued strength; he stresses selecting the strongest ideas rather than buying every stock in a momentum category.
• He generally dislikes relying on old investing books because strategies and market conditions change. He prefers business histories, which can reveal useful lessons about leadership and execution.
• He recommends limiting exposure to online groupthink and finding ideas through genuine interests, then asking whether they create value and can make money.

Views on AI Investments

• He prefers the risk profile of large hyperscalers such as Google, Amazon, and Microsoft over companies more exposed to hardware demand swings.
• In his view, some compute contracts may protect hyperscalers through customer commitments, while customers and hardware suppliers could face greater downside if demand or pricing collapses.
• He had owned Taiwan Semiconductor and other AI-related stocks but reduced that exposure as valuations and downside risk changed.
• He says his portfolio’s correlation to the AI factor was approximately zero at the time of the interview.

Broader Business Plans

• The fund is exploring an internal stock-borrowing operation to reduce short-selling costs, which the guest estimates could already save around $1 million annually.
• He says fundraising has been relatively easy, but his priority is improving the net-of-fees track record, not maximizing assets under management.
• His view on developing an edge is to keep learning new tools—especially AI—quickly and practically, rather than trying to predict which capabilities will remain defensible over the long term.