Yields.io – An AI Platform For Model Risk Management
I interviewed Jos Gheerardyn, Co-Founder & CEO, and Sebastien Viguie, Co-Founder & CTO at Yields.io. Yields.io is a Belgian Fintech Startups on a mission to empower quants, risk managers and model validators with smarter tools to turn model risk into a business driver.
Yields.io – An AI Platform For Model Risk Management.
Why did you create Yields.io?
Over the past 15 years Jos Gheerardyn and Sebastien Viguie have been creating mathematical models in the financial and energy sectors. These models are used in high risk environments, meaning that the decisions that are made with these algorithms have a very high business impact and can literally change people’s lives. Because of this, the models used cannot fail (meaning that e.g. we had to give a 99.999% SLA on the algorithm). It turned out that although many solutions exist to test software automatically, no such tools can be found to test mathematical algorithms.
This is why we created Yields.io.
Founder personal story
Jos Gheerardyn and Sebastien Viguie built Chiron, the first FinTech platform that leverages AI for real-time model testing and validation on an enterprise-wide scale. MRM is a zealous proponent of model risk governance & strategy.
Jos is the co-founder and CEO of Yields.io. He is on a mission to empower quants, risk managers and model validators with smarter tools to turn model risk into a business driver. Prior to his current role, he has been active in quantitative finance both as a manager and as an analyst. Over the past 15 years, he has been working with leading international investment banks as well as with award-winning start-up companies. Jos is the author of multiple patents applying quantitative risk management techniques to imbalance markets and holds a PhD in superstring theory from the University of Leuven, Belgium.
Sebastien is the co-founder and CTO of Yields.io. He is a senior quantitative analyst professional with extensive knowledge of mathematical modelling. He received his M.Sc. in financial mathematics from Stanford University in 2006 before spending more than 10 years developing his expertise between Brussels and London within the BNP Paribas Front Office research team. Knowledge of various programming languages and software technologies, Sébastien concentrates on the product technical development of Chiron, Yields.io platform, from the Belgian offices.
What is your vision?
Due to the abundance of machine learning solution providers and the introduction of AutoML (i.e. software tools to create machine learning algorithms automatically), the very fact of using mathematical algorithms (and especially machine learning) will soon cease to be a differentiating factor for many businesses.
What will matter is how institutions manage these mathematical models. Only with proper governance will an institution be able to quickly evolve its modelling approach in order to adapt to changing market trends and to capture the value of mathematical models sustainably.
This evolution parallels what happened in software development already few decades ago and will transform model risk management from a cost center into a value driver.
How do you help your customers and partners?
Financial institutions use a large and ever-growing number of mathematical models to conduct their business. A typical regional bank has around 100 models in production while the large Tier 1 investment banks have over 3000 algorithms and that number increases by approximately 15% per year due to new regulatory frameworks as well as the introduction of AI applications. Banks therefore need to scale their model risk management processes to deal with model risk challenges.
Yields.io offers a comprehensive model-risk platform to allow financial institutions to automate and industrialize model validation and monitoring. This allows our users to address model risk in a cost-efficient and future-proof fashion. Our clients use our solution to guarantee compliance with regulatory initiatives and international standards such as TRIM, IFRS 9, BCBS 350, PS 7/18, SR 11-7 and others.
Innovation and Artificial Intelligence together.
Yields.io leverages both technological innovations as well as AI to help our clients to turn model risk management into a business driver. The solution has three components:
- A data governance module that centralizes all data needed for model risk in a distributed data lake. The solution supports data lineage, data versioning and the measurement and analysis of data quality.
- The analysis module which contains an interactive workspace for validators to deep dive into the model, measuring model performance using numerous statistics and comparing the model with automatically generated benchmarks. Registered analysis can be run continuously to support ongoing monitoring and detect issues in real time.
- A reporting module generating both technical validation documents for risk managers and audit, as well as creating interactive dashboards for C-level executives
Financial institutions use Yields.io across the three lines of defence.
First-line uses cases include so-called pre-validation, i.e. automated model testing and report generation. Another common use case is ongoing monitoring of e.g. valuation and XVA models. Second line applications are mostly centred around automating and industrializing the validation process. Third line applications are the creation of additional challenges.
Yields.io is primarily a software company offering our solution both on premise as well as on the cloud (SaaS).
What makes Yields.io different?
Yields.io is the only comprehensive and AI driven model-risk management platform. Competition focuses only on part of the problem:
- Consulting companies offer manual validation services. Chiron automates this process therefore speed up the validation and decrease costs.
- With Chiron, any institutions make a strategic commitment to own the validation process as opposed to outsourcing it.
- Self-service analytics platforms focus on benchmarking solely, i.e. the generation of alternative models to study the impact of changing model assumptions. However, a large part of model risk is related to analysing data quality and industrializing the process, which is present in our platform but missing in those vendor solutions. This is why Yields.io can generate model validation documents compliant with the strictest regulatory frameworks such as SR 11-7 and PS 7/18.
- Reuters or SAS model inventory solutions address the workflow of a validation process but do not have the quantitative information on real-time model issues.
- Companies such as Moody’s Analytics or James Finance focus primarily on a single type of model (credit risk) while our solution is generic with respect to the model types. In addition, Chiron enable to analyse models that have not been developed in its framework.
How to get in contact with Yields.io?
Customers can contact us through the website – Yields.io
Would you like also to include an article about your Tech Startup and to be featured on our Tech Startup Directory you can contact us by email or write a comment below and we will contact you.
Javier Nieto León