Managing model and AI risks in the investment management sector

Understanding and mitigating model risk

The investment management industry is increasingly using models and complex algorithms to enhance the speed and accuracy of decision-making. Hence, investment managers are becoming more interested in comprehending the sources of model risk and standing up model risk management frameworks—while also considering new risks associated with generative AI models and digital assets.

Model risk management: What’s driving investment managers?

Models are pervasive in the investment management industry, and they are used to facilitate important business activities, such as asset allocation, algorithmic trading and portfolio rebalancing, market and liquidity risk management, and regulatory compliance. While models help organizations drive competitive advantage and achieve operational efficiencies, not managing them effectively can lead to flawed predictions and erroneous decisions—which in turn can erode investor trust and damage the reputation of investment firms.

Given the increasing attention to models by investors, stakeholders, and regulators, investment managers need to proactively design an effective model risk management framework to mitigate the strategic, regulatory, and operational risks for businesses.

What is model risk, and why model risk management?

Model risk can be understood as the risk of experiencing monetary loss, harm to clients, erroneous performance or risk metrics, improper investment or managerial decisions, or damaged reputation resulting from poorly built, used, or controlled models. To mitigate the model risk, model risk management (MRM) is a discipline of risk management that provides a structured approach across the model life cycle. It can help organizations define the shared roles, responsibilities, and accountabilities (inclusive of decision rights) across business functions, and facilitate the development of an effective control environment, including policies, procedures, and corollary controls.

Leading practices in enhancing model resilience

Rigorous model development processes, a broad model testing and evaluation approach, and an effective model operations framework serve as the cornerstone for an organization’s robust model environment. We explore leading practices to enhance model resiliency with a focused lens on artificial intelligence (AI) models.

Future of Work: Ways of working in uncertain times

Rigorous model development

The model development process should include a rigorous set of steps to solve a problem using data and analytics. The leading practices in the industry place a strong emphasis on the problem definition as a first step to fully understand the business issue.

Systematic and independent model validation

Model validation plays a crucial role for investment managers in determining the accuracy, reliability, and appropriateness of the models and their use in the investment decision-making processes.

Efficient model operations: ModelOps and machine learning operations (MLOps)

Expanding on our understanding of leading practices, MLOps is an engineering framework of tools and practices used to streamline and manage the lifecycle of machine learning (ML) models in the production environment.



Managing emerging risks related to generative AI

Organizations that aim to manage generative AI risks should consider beginning by managing the risks already identified with “traditional” AI. These risks can be mitigated by addressing model risks such as the potential for bias in data or models, or lack of accuracy of the output. This is in addition to ethical considerations, data privacy, and safety issues. Investment firms need to determine the proper oversight, validation, and monitoring of generative AI systems to maintain transparency, fairness, and accountability in their operations.

By embracing generative AI technologies responsibly, investment management firms can gain a competitive edge, provide more value to clients, and adapt to the evolving landscape of the industry.

Managing model risk related to digital assets

As institutional interest in investing in digital assets continues to rise, there are additional financial risk management challenges to consider. Market risk models designed to evaluate the risks and returns of traditional financial assets do not address the idiosyncrasies of risk factors of cryptocurrency and digital assets as an alternative asset type. In addition, limited liquidity and fragmented markets for certain cryptos with smaller market size make valuation methodologies inadequate to estimate the market value for instruments based on digital assets.

Now is the right time

By recognizing the importance of model risk management and taking appropriate actions, investment managers can navigate the complexities of the industry, adapt to changing market dynamics, and strive for sustainable long-term achievements. Download our report to learn more.

Get in touch

Krissy Davis
US IM Practice Leader | Risk &
Financial Advisory
Deloitte & Touche LLP

Clifford Goss, PhD
Partner | Risk & Financial Advisory
Deloitte & Touche LLP

Anca Ferent
Senior Manager | Risk & Financial Advisory
Deloitte & Touche LLP


Dan Han
Manager | Risk & Financial Advisory
Deloitte & Touche LLP

EJ Donelson
Analyst | Risk & Financial Advisory
Deloitte & Touche LLP

James Stanley
Consultant | Risk & Financial Advisory
Deloitte & Touche LLP

Palak Kaur
Senior Consultant | Risk & Financial Advisory
Deloitte & Touche Assurance & Enterprise Risk Services India Private Limited

Contact us

Connect with us for more insights to help you adapt to regulatory developments.

  Yes         No

Fullwidth SCC. Do not delete! This box/component contains JavaScript that is needed on this page. This message will not be visible when page is activated.

Did you find this useful?