Alternative Market Briefing

How AI is changing the hedge fund investment processes

Wednesday, October 21, 2026

Artificial intelligence could be one of the most significant developments in hedge fund investing since the emergence of quantitative trading, according to Don Steinbrugge, founder and CEO of Agecroft Partners, a global hedge fund consulting and marketing firm.

What sets this wave apart is that it is not confined to quant funds with teams of PhDs, he explains in a recent paper. Generative AI and machine-learning tools now give discretionary equity managers, credit specialists, macro traders and small funds access to advanced research capabilities. AI lets investment professionals analyse far more information, faster and at lower cost.

Six areas stand out.

1. Fundamental research

Analysts used to spend days reading filings, earnings releases, call transcripts and competitor reports, mostly locating and organizing information rather than analysing it. AI can now:

AI can also map supply chains, summarize expert-network notes, and read foreign-language filings, widening the markets a team can cover.

But AI can also misread tables, cite outdated data or deliver wrong answers with confidence. Analysts must verify key figures, and firms need policies on data security, material non-public information and documentation.

Because these tools are widely available, access alone is no lasting edge; advantage will come from how firms integrate AI, their proprietary data and their judgment.

2. Alternative data

AI makes it far cheaper to analyse unstructured data such as satellite images, card transactions, shipping activity, job postings, reviews, web traffic and social media. It can count cars in parking lots, read millions of reviews, and align datasets from different vendors. For a retailer, combining several sources could produce a revenue view before earnings and show whether growth stems from new customers, higher spending or discounting.

The advantage lies in combining datasets, not buying them. Single sources can mislead, since card panels skew demographically and satellite data is affected by weather. Risks include short data histories, shifting vendor methodologies and privacy rules, so vendor due diligence matters. AI may make such data accessible to smaller managers, but as more funds use the same inputs, signals will be priced in faster and decay sooner.

3. Quantitative strategies

Machine learning can detect nonlinear relationships that traditional factor models miss, and natural language processing turns text into signals. Applications include identifying market regimes, forecasting volatility, combining weak signals into stronger ones and adjusting risk as correlations shift.

This is also where AI is most dangerous. Markets have low signal-to-noise ratios and evolving relationships, so complex models can fit noise and produce impressive back tests that fail live. Look-ahead bias, opaque models and crowding add risk. Winners will be those with the best data, research discipline and out-of-sample validation, an economic rationale for each signal, realistic cost assumptions, and monitoring for decay.

4. Portfolio construction and risk

Conventional risk systems focus on sectors, countries and standard factors, and can miss true economic drivers. A portfolio holding a chipmaker, a utility, an electrical equipment maker, a commercial real-estate developer and a copper miner spans five sectors, yet all may depend on AI data-centre spending. AI can uncover such hidden exposures by analysing fundamentals, supply chains and trading behaviour.

Other uses include stress-testing unprecedented scenarios, forecasting liquidity against redemption terms, tracking correlations that spike in crises, scenario modelling, and spotting crowded positions. For multi-strategy funds, AI can aggregate exposures across dozens of teams almost in real time. It should support, not replace, risk judgment, since models trained on history may underestimate unprecedented events and unexplainable outputs are hard for risk committees to act on.

5. Trade execution

Transaction costs include spreads, market impact and opportunity cost, and they directly reduce returns. AI can predict liquidity, select algorithms and brokers from a fund's own history, estimate market impact before trading, optimize timing, and detect information leakage. It also helps in less liquid fixed-income and credit markets by estimating fair value and pricing quote requests.

The math is compelling: a five-basis-point (0.05%) saving on a $500 million purchase is $250,000. For a $5 billion fund trading $15 billion annually, it is about $7.5 million, or roughly 15 basis points of added return. Large, high-turnover and illiquid-strategy managers benefit most, while smaller firms will rely on broker and vendor tools. Execution is becoming an arms race in which today's edge quickly becomes tomorrow's standard.

6. AI agents and automated workflows

This may be the most transformative development over the next three to five years. Unlike a tool that answers one question at a time, an agent takes an objective, breaks it into steps, retrieves data, runs analyses, checks results and decides what to investigate next, much like a junior analyst.

A typical workflow: a human defines a hypothesis (say, data-centre power demand benefiting certain utilities); the agent researches exposed companies, filings, competitors, alternative data and scenarios; it delivers a sourced report with models, risks and catalysts; and a portfolio manager reviews and decides. Agents can also monitor positions and alert teams when new information challenges a thesis.

Automated research is not autonomous investing. Early errors can propagate, so firms need limits on agent actions, full audit trails and human sign-off. Small teams may become far more productive, and fewer analysts may be needed for gathering information. If many firms use similar agents, conclusions may converge, making independent thinking more valuable.

Implications for managers

Large funds have historically led in staffing, technology and data. AI could narrow that gap through subscription-based models and cloud computing, letting a five-person team do work that once needed many more. It can also cut compliance, reporting, due diligence and marketing costs, lowering the assets needed to run a viable business and helping emerging managers. Smaller firms may adapt faster and find opportunities in niche, capacity-constrained strategies.

Large firms still hold advantages in exclusive data, computing infrastructure, scarce AI talent and the ability to experiment. As public-information analysis becomes commoditized, proprietary data gains value. AI will narrow the gap in some areas and widen it in others.

Implications for allocators

The key due diligence question is not whether a manager uses AI but whether it genuinely improves the process and whether risks are controlled.

Allocators should ask how AI is integrated into decisions; what controls govern data validation, material non-public information and agent limits; which alternative datasets are used and what the manager does with them that rivals do not; how systematic managers guard against overfitting and compare live results with back tests; and how risk exposures, stress tests and transaction costs are managed. Allocators can also apply AI to their own portfolios to find hidden overlap and crowding across managers.

Conclusion

AI is making information cheaper to gather and analyse, so public-data signals will be priced in faster and research that took a week may take a day. Firms that do nothing risk falling behind; those that move too fast risk trading on hallucinations or overfit back tests.

What will separate managers is what AI cannot supply: disciplined validation, exclusive data and sound judgment. As firm size becomes a weaker proxy for capability, process quality will matter more, and allocators must understand how managers use AI and what controls limit the added risks.

Article link