SimCorp and Axyon AI have jointly published a white paper examining whether AI-generated stock-ranking signals, when combined with institutional-grade portfolio optimisation, can deliver consistent active returns. The research, From AI Signals to Active Returns: A Portfolio Manager’s Guide to Capturing Consistent Alpha, analyses a 10-year backtest period from 2015 to 2025 across US All Cap equities benchmarked against the Russell 3000 index.
The headline findings are respectable: long-only strategies using Axyon AI’s predictive signals and Axioma by SimCorp’s optimisation and risk modelling tools generated positive active returns in 61 to 65 percent of all months tested, with the majority of the alpha attributed to stock-specific selection rather than factor tilts.
But for institutional data and analytics professionals, the more significant story may not be the backtest results themselves. It is what the white paper reveals about SimCorp’s evolving approach to packaging and distributing its analytics intellectual property, a strategy that has accelerated markedly in 2026.
From Risk Models to Research Inputs to Partner Signals
The white paper is the latest in a sequence of moves through which SimCorp has been systematically opening up the analytics layer behind its Axioma risk models.
In March 2026, SimCorp launched the Axioma Factor Library Suite, making a broad set of proprietary equity and macro factors – previously generated internally as part of Axioma’s risk model research – available as a standalone dataset for quant investors. Distributed via the Snowflake cloud platform, the Factor Library was explicitly positioned as research inputs rather than pre-packaged strategies: raw building blocks for hedge funds and systematic investors to incorporate into their own pipelines.“We generate these factors daily as part of our modelling infrastructure, but many were never previously distributed,” Ian Lumb, Head of Analytics Product Management at SimCorp, told Market & Alt Data Insight at the time. “The new library gives clients access to that broader research set so they can apply and customise the signals within their own investment frameworks.”
The Axyon AI white paper extends that logic in a different direction. Where the Factor Library serves quant teams who want to run their own research, the Axyon collaboration demonstrates what happens when a third-party AI signal provider is pre-integrated into SimCorp’s portfolio construction and risk infrastructure, and provides the backtest evidence to support adoption.
The Partner Ecosystem as a Distribution Channel
Axyon AI, an Italian fintech specialising in predictive AI for asset managers, joined SimCorp’s curated partner ecosystem in September 2025. The partnership integrates Axyon’s deep learning-based stock-ranking signals directly into the SimCorp One platform, giving portfolio managers access to predictive analytics within their existing workflows without requiring a separate IT integration project.
Anders Kirkeby, Head of Open Innovation at SimCorp, has described the partner ecosystem model as one designed to de-risk technology adoption. “When we add partners to our programme, we take care of the integration so the client doesn’t need to worry about it,” he told A-Team Insight when the partnership was announced. “You can run a backtest against past performance and then use it in a real-world scenario on your own portfolio and see how it works.”
The white paper is, in effect, that backtest made public. Daniele Grassi, co-founder and CEO of Axyon AI, says in the paper that the research demonstrates Axyon’s signals serve as a consistent source of active return within a broader US equities mandate. Melissa Brown, Head of Investment Decision Research at SimCorp, frames the combination of Axyon’s signals with Axioma’s optimisation capabilities as a means of ensuring that alpha forecasts translate into consistent, risk-adjusted portfolio performance, rather than being eroded by poor portfolio construction.What the Research Shows. And What It Doesn’t
The white paper tests a range of long-only strategies at different active risk targets, designed to assess whether AI-driven signals can generate outperformance across a full market cycle. The results suggest the approach may produce active returns that are broad-based and repeatable, with the strategy delivering positive months in the majority of cases over the decade tested.
The claim that the alpha is predominantly stock-specific is potentially significant. Many AI-driven strategies, when examined closely, turn out to be repackaging familiar factor exposures – value, momentum, quality – in a new wrapper. If Axyon’s models are genuinely picking individual winners and losers based on patterns that fall outside those established categories, that would make the signal more useful to portfolio managers who are already harvesting returns from traditional factors and looking for something additive.
However, the research leaves several questions open that institutional investors would reasonably want answered before acting on the findings.
The entire analysis is based on historical backtesting. There is no claim of live trading performance, and the white paper does not address the well-documented challenges of translating backtested equity strategies into real-world execution, including transaction costs at scale, market impact, and the potential for overfitting in model selection.
The research is also silent on the nature of Axyon AI’s underlying models. While Axyon describes its technology as an AutoML platform that automatically selects algorithms and datasets for prediction tasks, the white paper does not disclose what data inputs the models consume, how frequently they are retrained, or how the firm manages the risk of model degradation over time. For quant researchers evaluating the signal, that opacity may be a limiting factor.
The Crowding Question
One issue that surfaces naturally when a platform vendor distributes AI signals through an ecosystem serving hundreds of institutional clients is the potential for crowding. If the same signals are consumed by multiple portfolio managers using the same optimiser, strategies could begin to converge, potentially compressing the very alpha the white paper documents.
Lumb addressed a version of this question in relation to the Axioma Factor Library, arguing that the library provides shared research inputs rather than a shared strategy framework, and that institutional investors typically combine factors differently enough to produce diverse portfolio outcomes.
The Axyon AI white paper presents a tighter test of that argument, because it demonstrates a specific signal-plus-optimiser combination with specific performance characteristics. The more widely that combination is adopted within the SimCorp client base, the more relevant the crowding question becomes.
Competing Platform Models
The SimCorp approach – curated partners, pre-integrated signals, evidence-backed adoption – represents one model for how platform vendors can help the buy side operationalise AI. But it is not the only one emerging in the market.
S&P Global has taken a structurally different approach with its Capital IQ Pro platform, exposing data and capabilities through Model Context Protocol (MCP) servers so that clients and their AI agents can build their own workflows on top of S&P’s data estate. In that model, the platform is open infrastructure, and the client controls how data is assembled and consumed. An enterprise pricing model is designed to reduce friction across the entire dataset.
SimCorp’s model inverts that relationship. The platform vendor selects and integrates the partners, provides the evidence base through research like the Axyon white paper, and presents it to clients as a curated, lower-risk path to adoption. The trade-off is that clients operate within SimCorp’s framework rather than building independently on open infrastructure.
Both approaches are credible responses to the same underlying challenge: the 2025 Global InvestOps Report found that 75 percent of buy-side executives recognised AI’s potential but needed more guidance on practical application. SimCorp is answering that with curation and evidence. S&P Global is answering it with openness and infrastructure. The buy side will ultimately determine which model – or which combination of models – delivers more value.
The Bigger Picture
Taken together, the Axioma Factor Library and the Axyon AI white paper suggest SimCorp is pursuing a dual strategy: serving quant teams who want raw research inputs they can customise, while simultaneously offering portfolio managers a turnkey, pre-validated AI signal pipeline embedded in their existing workflow.
The white paper’s backtest results provide a useful data point, but the more durable story is the strategic one. SimCorp is systematically monetising the analytics intellectual property generated by its Axioma research infrastructure, first as standalone data products, and now as validated signal-plus-optimiser combinations delivered through its partner ecosystem. Whether that model scales without diluting the signals it distributes is a question the next phase of adoption will need to answer.
Subscribe to our newsletter
