Quant Foundry Model Suite – High Level Descriptions
AI/ML core engine (multiple use cases)
The core engine provides the analytical framework for the appropriate and safe deployment of mostly AI/ML techniques. The core engine utilises several sources of transactional, valuation input, reference, user-input data and process taxonomy across the KRM 22 platform to provide immediate and targeted insight. Example models include recommendation engines to improve customer engagement, product engagement models.
Operational Risk Smart Monitoring
A machine learning model that monitors realised loss data, process taxonomy and then defines potential influencers such as resource shortfall that lead to operational risk events. These are used to provide immediate insight into operational pain points that, when remediated, can contribute to a lower risk profile as potentially a lower capital charge. Model is designed to improve operational risk forecasting, providing a means for companies to assess and categorise risk with enriched datasets incorporating bespoke factors. The model can also be trained across multiple clients data and would provide insight learned across clients.
The model allows companies to build bespoke risk classifications (or taxonomies) as well as a standard template then, when used across multiple clients provides peer benchmarking. The model integrates with the Operational Risk Capital Engine.
Operational Risk Capital Engine
The Operational Risk Capital Engine operates either standalone or fully integrated with the Operational Risk Monitoring Model. The model permits the user to calculate operational capital calculations using a combination of different methods utilising Standard approaches, Loss Distribution Approach, quantitative scenarios, including stress testing and insurance mitigation.
Stress Scenario Indicator
Stress Scenario Indicator provides a distribution of a company’s potential return and volatility, providing insight into the likelihood of a paradigm shift. The model learns from previous behaviours where the share price has boomed to then collapsed. The model highlights clusters of over-valued companies whose “bubble anatomy” can be mapped to historical instances. The model can then define the likelihood and severity of each bubble bursting in the next 12 months.
Coherent Risk Scenario Generator
The model helps an organisation to construct a coherent suite of market and counterparty risk scenarios. The model enables users to construct a coherent view of market impacts across a broad set of risk factors from a narrow set of baseline stress scenarios. The model uses a combination of market and macro-economic indicators and then permits users to build stress scenarios across multiple risk factors.
Network Resilience model
Network model that measures the exposure and liquidity profile of the leading players that are exposed to a bubble bursting. The model simulates the initial shock and the immediate response of reducing exposure during the period of downturn. Measures the overall resilience of the players – e.g. capital buffer – to absorb initial losses. The model measures the potential for a sell-off in other related assets leading in-extrema to a full contagion experienced in 2008. The model can be used standalone or integrated into the stress scenario indicator and scenario generator packages.
Climate Transition Risk
The model measures the impact on a company’s balance sheet as a result of their exposure to the effects of climate change and related government policies. The model covers the increased physical risk of warming due to a change in severity and frequency of weather events as well as the cost of transitioning production and consumption to a carbon-free paradigm.
Sovereign and low probability issuer default
A Machine Learning tool that breaks down the thought process of a credit analyst to provide an unbiased, stable and superior default predictor that enables the investor in government bonds to assess the probability of rating migration over a one-year horizon. The model learns from history in terms of the complex combination of internal and economic conditions that preclude a rating migration and builds a suite of risk factors that contribute to the accuracy and the explanatory power.