01
Bring AI into your investment process.
You have a strategy or an investment view. We turn it into an AI-driven, risk-managed, executable strategy, built around your rules and validated against real market conditions.
INSTITUTIONAL-GRADE AI FOR QUANTITATIVE INVESTING
Quantitative AI, built for institutional investors.
We are your AI and technology partner. We build deep learning models for financial time series and portfolio construction, designed around your strategy, your rules and your regulatory requirements.
Book an Introductory CallWhat You Get
What it does for your firm.
Four things change when your AI work runs on a framework that is already built and already in production.
01
Speed
Your edge reaches the market faster, because the framework is already built and tested in live conditions.
02
Control
Models your team fully owns and understands, with the source code licensed to you and no dependency on us once the work is done.
03
Standards
Models validated out of sample, explainable, and governed to the standard your risk, audit and regulatory functions already require.
04
Focus
Your team stays focused on the strategy, while the AI engineering runs on a foundation that already works.
One Kind of AI
One kind of AI. Built for markets.
Built for market data
The AI we build.
Deep learning models trained on market data, built for financial time series, and integrated into portfolio construction and risk management. Explainable, validated out of sample, and governed like any institutional system.
Not built for markets
We do not use language models.
A language model learns patterns from text that has already been written. It is built for work that repeats itself. Markets do not. It carries no understanding of price, liquidity or risk, and no amount of prompting makes it a market model. So it is not part of anything we build, and no number in our work comes from one.
The pattern it repeats is never the one in front of it.
A model built on repetition eventually meets a situation it has not seen, and it has no way to know when an answer is wrong. That is why we build our own models for financial time series, and why no number we publish and no trade we take is decided by a language model. That is not a rule we chose. It is a limit of the technology itself.
Almost every investment firm now uses AI. Few use it to improve investment decisions.
Most AI in investing goes into productivity: summarizing research, drafting notes, searching documents. That is useful, but it is not where investment decisions are made.
Bringing AI into the investment process itself takes something different: models designed for financial time series, built into portfolio construction and risk, and engineered to the standard an institution needs. Most firms do not have a dedicated team to build that.
That is the gap we close. We bring the AI expertise and a framework that is already built, and you keep your strategy, your rules and the code.
01
SINCE 2000
AI in markets
Our founder's first paper on AI in trading, Taylor & Francis
02
BUILT
Our own AI framework
Running live in weekly production
03
INSTITUTIONAL
30 years
Risk technology leadership at BNY Mellon, Deutsche Bank and Barclays
04
DELIVERY
Full ownership
Source code licensed to every client
Where We Help
Wherever you are with AI, we meet you there.
02
Strengthen the AI you already have.
You have a quant or AI team. We add models, data, engineering capacity and production infrastructure, or plug your own models into our framework, so your team moves faster.
03
Move AI from research to production.
You have models that work in research. We make them robust, repeatable, auditable and connected to how your portfolio is actually run.
04
Trust the AI you rely on.
An independent review of your models for data leakage, overfitting and fragility, including explainable-AI audits of each decision.
Pouthon Eigen
Your strategy runs on a framework that is already tested.
Pouthon Eigen is our technology framework, built and tested in live market conditions. It brings our own AI models together with your investment logic, directional signals, portfolio allocation, risk management, and execution in one consistent system.
When you work with us, your strategy does not start from a blank page. It starts from a working foundation, developed by someone who spent 30 years building exactly this kind of infrastructure inside the world's largest banks, configured around your ideas and your constraints. You keep the code.
Directional Signal Layer
Per-security directional signals across multiple time horizons, fed by macroeconomics, credit and treasury yields, company fundamentals, news alerts, options volatility surfaces, and peer group correlations.
Why Pouthon
A head start, not a blank page
Eigen is already built and tested. Client projects begin from a working framework, reducing time, cost, and delivery risk significantly compared to building from scratch.
We speak your language
We hold a credible conversation with your CIO, your head of risk, and your lead engineer in the same meeting. Fewer translation errors between investment and technology teams means faster, better outcomes.
AI built for markets
We have applied AI to trading since 2000. Our own models are designed for financial time series, not borrowed from other domains. Security-level signals, regime-aware overlays, walk-forward validation, and no data leakage help clients build on a tested foundation.
You own the result
Every engagement ends with a full license to the Eigen code used in your project. No vendor lock-in. Your team can extend it independently, or with anyone else you choose.
The reason firms hire Pouthon.
Most systematic strategy firms are built by data scientists who have never sat on a trading floor, or technologists who have never managed risk with real money on the line.
Arthur Rabatin has done both for 30 years, at the senior leadership level, inside the institutions that define global finance.
As Head of Markets Risk Technology at BNY Mellon. As Head of Counterparty Credit and Funding Risk Technology at Deutsche Bank. As Head of Credit Correlation and Exotics Technology at Barclays Capital. Building the systems that trading desks relied on. Understanding what breaks under real market conditions. Knowing what governance, audit, and regulatory scrutiny actually require.
He published his first paper on AI in trading in 2000. This is not a new interest. It is a 25-year body of work applied to a very specific problem.
When you engage Pouthon you are not hiring a consultancy that will figure out your world. You are hiring someone who has already lived in it.
BNY Mellon
2016 – 2023
Deutsche Bank
2010 – 2016
Barclays Capital
2005 – 2010
EdF Trading
2002 – 2005