1. Could you please introduce 3AI and share your company’s mission?
3AI is an artificial intelligence (AI)–driven investment intelligence company focused on systematically identifying and supplying forward-looking alpha to the investment industry.
We describe 3AI as an alpha refinery: a new type of business that combines large, complex datasets with deep subject matter expertise in investing and proprietary machine learning to mine forecast investment insight. That refined alpha is delivered in scalable, usable forms, including indices.
3AI originated from an award-winning data science team whose self-learning investment AI achieved national recognition in the U.K. in 2017. The technology was deployed to trade billions of dollars and generated insights adopted by the chief investment officer of a large insurer. At that time, we rejected a proposal from the insurer to spin out, choosing instead to form 3AI.
The team combines decades of institutional asset management experience with deep expertise in machine learning and applied data science. Over time, the system repeatedly demonstrated an ability to forecast stocks with accuracy materially greater than would be expected by chance.
2. In your view, how is AI particularly relevant and ready for use within indices today?
Building on the first wave of automation introduced by indices, machine learning (a subset of AI) represents the next natural evolution in systematic investing. Traditional beta indices democratized access to broad market returns by offering low-cost, transparent solutions. Factor indices further advanced this by providing systematic, rules-based exposure to specific drivers of return, enabling more intentional portfolio construction. Today, machine learning enables a much deeper examination of factors or financial data. By learning from history and applying that learning to current information, AI models can generate forward-looking perspectives. At 3AI, we refer to this as Alpha Intelligence. In essence, we use machine learning to compress deep factor understanding into a single intelligent factor, i.e. a model that adapts to market conditions or stocks to craft a bespoke factor framework that compresses hundreds of factors into the one we really care about: alpha, or in other words, what could happen tomorrow.
We also believe that machine learning can be leveraged in investing to address core challenges such as model risk, overfitting, noisy and complex datasets, feature selection bias and point-in-time integrity. Advances in computing, data availability and algorithms now make this approach practical scalable, and robust for index applications.
Importantly, modern AI systems can also provide meaningful explainability. At 3AI, forecast alpha can be attributed to individual data inputs. Evidence from live markets, including indices tracking hedge funds, shows that AI-based approaches already outpace peers in both absolute and risk-adjusted returns even at this early stage. As data availability expands alongside algorithm and computing power advancements, the capabilities of AI are set to accelerate further. Together, these developments are powering an exponential surge in the power of AI.
3. What is machine learning, and what specific advantages, in your view, does it offer compared to traditional human-led data analysis?
Machine learning is a step change in automation, but it is fundamentally an extension of human intelligence, not a replacement. At 3AI, all systems are human designed and human governed, and our Alpha Intelligence is therefore deeply human led.
The key difference versus traditional analysis is that machine learning allows problems to be solved systematically and at scale,rather than repeatedly by humans. Algorithms encode humanity’s best ideas about learning and optimization, amplified by computing to operate at superhuman speed and scale.
Investing is one of the most complex decision-making challenges that exists, involving uncertainty, vast choice sets and long-dated path dependencies. Machine learning offers clear advantages here, including the ability to retain and analyze vast histories of observed data, learn empirically without cognitive biases, operate at superhuman research speed, and detect complex interdependencies across markets.
At 3AI, this is complemented by a “learning about learning” layer that continuously evaluates how well models are performing in real time. Models must continually prove themselves to remain active, ensuring discipline and governance.
