AI is reshaping markets. It's also changing how investors find opportunity.
Expert Perspective
|November 12, 2026
Expert Perspective
|November 12, 2026
Quantitative Equity (QE) is Vanguard's in-house active management team, operating as a nimble boutique backed by the firm's scale and resources. The 28-person team manages about $60.2 billion1 and has been refining its approach since 1991, working directly with data scientists from Vanguard's AI and machine learning group. In this conversation, QE head Cesar Orosco discusses how the rise of AI is shining new light on that time-tested approach, where human judgment still matters, and what it means for advisors.
AI is influencing our work in two ways. The first is speed: Faster processing lets us cover much more ground, more rapidly. The second is depth: These tools can help us analyze more information and uncover patterns and relationships that can strengthen our investment research. The tools may feel new to many people, but we've been using them in our investment process for years, long before they became mainstream. What AI hasn't changed is our philosophy. We believe companies with solid fundamentals, strong growth prospects, and market tailwinds are likely to be better investments. That can't be captured simply by throwing data at a machine and letting it do whatever it wants; you have to impart an understanding of the economy. Inside our models, we nudge the machine in the direction we know it needs to go. AI has given us more powerful tools to build that philosophy into the process.
Traditional active managers often build a thesis around a handful of favorite stocks. That's a valid approach, but it isn't ours. We model the characteristics that give a portfolio a good chance of outperforming. Because of our diversification and risk controls, we never bet heavily on any single company. We bet very strongly on the model.
"We don't bet heavily on any one company. We bet very strongly on the model."
Cesar Orosco
Head of Vanguard Quantitative Equity
Vanguard's Indexing team and our team are two branches of the same tree. Both began in the same group at Vanguard, so we bring the same strengths advisors expect from indexing, including rigorous risk controls, broad diversification, and low costs, to active management.
We're also built around a system, not a star. If any one person leaves, the model survives, and the researchers remain to keep advancing it. Our longest-running fund has followed the same philosophy for more than 30 years, evolving over time.
AI tools and computing power are easy and inexpensive to access. Good data is not.
Investing has a challenge many other fields don’t: the signal-to-noise ratio is extremely low. Much of the information that is freely available has very little real content, and if you feed that into a model, you end up finding noise. That's why the question isn't whether we have too much information, it's whether we have enough high-quality information.
We've invested in high-quality data that feeds directly into our strategies. This high-quality data combined with more than 30 years of experience helps us tell what's useful from what isn't and improves our forecasts.
In the 1980s, the math was there, but the data and the computing power weren’t. The 2010s brought an explosion of data availability, computing power kept increasing, and the semiconductor breakthrough eventually helped bring large language models into the mainstream.
We were using machine learning techniques even before that, including natural language processing (NLP), an early way of converting text into numbers that can be used in a model.
In 2019, with machine learning ramping up and computing power finally available, we felt we owed our clients a clear, evidence-based answer on whether we should use it. So, we began a project to see whether we could build a model that reflected Vanguard's philosophy using machine learning techniques. By embedding that philosophy in proprietary architecture, we found very good results. The resulting model launched in January 2023 and has been powering many of our funds ever since.
At the beginning, the middle, and the end.
At the beginning is research. Experienced people design the model, our interpretation of Vanguard's philosophy, and our performance objectives. In the middle is implementation, where machines and people work together to make sure the model is properly vetted and runs robustly. This is running real money. Portfolio construction follows clear rules for how portfolios are diversified, how exposures are limited, and how risk and trading costs are managed. That process generates a list of buys and sells.
At the end, a portfolio manager reviews the trades to confirm they make sense from a philosophical perspective and that nothing has broken. We very rarely override the model, and that's deliberate. But we apply common sense. During the meme-stock frenzy, when GameStop surged for reasons that had nothing to do with fundamentals, we didn't take a stand on the stock. We neutralized its risk by holding it at benchmark weight and kept our focus where the model is designed to add value. Steps like that are important but infrequent.
Much of today's conversation centers on generative AI and large language models (LLMs). Our approach isn't to open a chat box, ask a question about a stock, and act on the answer. There's nothing wrong with that, but it's not how we operate. We research, build a model that is robust, replicable, and testable, and then implement it.
Over the past year and a half, just as we did for our deep neural network model, we've built a dedicated generative AI capability. It connects text and other non-numerical information with the rest of our data, so LLM research can be run at scale and in a reproducible way.
That required solving real technical challenges. One example is look-ahead bias: how do you test whether NVIDIA was a good buy in 2020 when the model has already been trained on what happened to NVIDIA afterward? We addressed those challenges, and now, when LLM research produces ideas that improve the model, we can implement them with confidence.
The objective is long-term outperformance. Behind it sits our data and research, more than 30 years of experience, and Vanguard's low costs. They're investing in a disciplined process built around a system, not a star, and one that keeps improving as our research advances. The result is portfolios that are highly diversified, tightly risk-controlled, and true to the Vanguard philosophy investors are looking for.
Strategic Small-Cap Equity Fund (VSTCX)
Commodity Strategy Fund (VCMDX)
1 As of October 31, 2026
Disclosures
For more information about Vanguard funds or Vanguard ETFs, visit advisors.vanguard.com to obtain a prospectus or, if available, a summary prospectus. Investment objectives, risks, charges, expenses, and other important information are contained in the prospectus; read and consider it carefully before investing.
Vanguard ETF Shares are not redeemable with the issuing Fund other than in very large aggregations worth millions of dollars. Instead, investors must buy and sell Vanguard ETF Shares in the secondary market and hold those shares in a brokerage account. In doing so, the investor may incur brokerage commissions and may pay more than net asset value when buying and receive less than net asset value when selling.
All investing is subject to risk, including possible loss of principal. Diversification does not ensure a profit or protect against a loss.
Prices of mid- and small-cap stocks often fluctuate more than those of large-company stocks.
The Factor Funds are subject to investment style risk, which is the chance that returns from the types of stocks in which a Factor Fund invests will trail returns from U.S. stock markets. The Factor Funds are also subject to manager risk, which is the chance that poor security selection will cause a Factor Fund to underperform its relevant benchmark or other funds with a similar investment objective, and sector risk, which is the chance that significant problems will affect a particular sector in which a Factor Fund invests, or that returns from that sector will trail returns from the overall stock market.
This article is listed under