“As the product lead for MSCI, I am naturally biased, but I believe the AI is transforming at MSCI the analysis of risk. Using the improving power of generative AI, we can start to measure how events are propagating through markets. Traditional factor models and time series analyses do a great job of capturing correlation, but now we can begin to capture causation as well,” says San Francisco-based Jay Dermody, Head of MSCI ONE, whose product management team helps provide actionable risk and performance insights for clients.

360-Degree Risk Management: A Necessity for Today’s Risk Leaders

The risk management function is evolving in ways that demand new infrastructure and technology to provide a 360-degree view of risk and informed investment decisions. This enables the risk management function to move from a back office control and compliance activity to a more strategic role.

MSCI participated in HedgeNordic’s 2024 roundtable on hedge fund operations, though its technology goes well beyond operational matters.

“Risk management must now be integrated into the end-to-end investment decision processes. This includes managing risks across multiple horizons, including macroeconomic trends to micro-level details and multiple spectrums, across market risk, credit risk, liquidity risk and inevitably climate risk,” says Sammi Shih-Yung Hsu, Head of Nordics Analytics Consultant at MSCI, who sits in London, UK. “Outdated infrastructure may struggle to handle the complexities and the large amount of data in a timely manner,” she continues.

Dermody underscores this point: “Risk management is increasing in complexity. We see this through the growth in alternative assets, including privates, and more demand for personalisation and customisation. At the same time, risk managers are increasingly asked to contribute as part of the investment team. We need to wrap our arms around this complexity with modern infrastructure, streamlined data collection, more specific models to capture hotspots and hedges, and a single view of risk across funds and portfolios.”

Smart Tech Tools and Solutions

Fortunately, technology can provide smart tools and solutions to elevate functionality and efficiency for both risk monitoring and advisory. Modern cloud technology, data warehousing and AI tools have the potential to provide better access to models and data, reduce errors in data-handling, and enable faster time to analytics. MSCI is continuously innovating to provide their clients with a better risk infrastructure. They work directly with hedge funds, collecting their data and leveraging it for the total portfolio across many dimensions and time horizons.

Get a Broad View of Risk and Return with Factor Models

Factor models are used by various investment managers to understand various sources of risks and return. MSCI also offers multi-factor risk models that help investors focus on factor exposures across asset classes in a consistent manner. “We specialize in data, content, factor models, and indexes. We started our model journey half a century ago, with the launch of Barra multi-factor risk models that are used to understand the investment characteristics that drive the risk and return of a portfolio via an integrated and consistent framework. Equities, fixed income, and privates are reported and we model how they all fit together into a portfolio and how each asset class and asset is likely to respond to market changes. We incorporate this into stress testing and scenario analysis,” explains Dermody.

MSCI’s pyramid-shaped framework goes from macroeconomic factors at the top, to position-level drivers of risk and return at the bottom. It can be sliced and diced by time frames and factors across multiple assets. The MSCI Multi-Asset-Class (MAC) factor model is a multi-dimensional approach encompassing factors from macroeconomic to single security levels, across asset classes like equity, fixed income, derivatives, and private assets. It supports various use cases such as risk reporting, portfolio management, and asset allocation, with timeframes ranging from one day to over five years. Many hedge funds use the tool for both risk management and portfolio construction across a variety of discretionary and systematic strategies. Earlier this year, HedgeNordic covered how MSCI analytics is enabling systematic credit and fixed income strategies in “The Systematic Revolution and ‘Equification’ of Fixed Income.”

Next Generation Factor Models

In addition to the traditional factor models based on quantitative data in financial statements, estimates, and datasets that are widely used, MSCI also offers next generation factor models that look at factors like sustainability, crowding, and machine learning, helping investors understand the impact of changing market conditions on their portfolios. “In the future, we will be able to provide richer analyses, such as understanding the impact of upside or downside surprise of a USA CPI release on a user’s portfolio,” points out Dermody.

All of this can be customised, personalised, and manipulated through interactive dashboards.

Case Study: March 2023 Mini-Banking Crisis

After the March 2023 mini-banking crisis, post-event analysis revealed crowding models had identified risks through equity factor models, correlations, and trends. MSCI Analytics models and solutions tracked risk factor performance and contagion between US national banks and US regional banks, and between EU banks and US banks. The models identified how regional banks started to move in a cluster based on peer similarity scores and correlations. The models also picked up some uptick in the volatility regime but did not find it was persistent. Models additionally identified tail risk based on five metrics used to gauge security crowding. All this informed users’ stress testing and sensitivity testing, flagging heatmaps needed for sector and stock selection.

Harnessing Snowflake

MSCI chose Snowflake for synthesising multiple datasets cohesively to provide a single source of truth for portfolios. “Snowflake is our provider of choice and it pushes content and client results into the environment. This removes the need for operationally complex traditional processes. Snowflake is superior to the traditional FTP flat file downloads that are still being used by some firms while others opt for APIs,” says Sammi Hsu.

The method works efficiently and provides compelling functionality for data, vendors, and clients. It graphs connections between MSCI content and clients or consumers and reduces data download timeframes from hours to seconds. “Snowflake can retrieve vast amounts of data very fast. It can pull the necessary information out of a table with 4.5 billion rows of datapoints in about 5 seconds. The datasets cover all asset classes, models, regions and countries, over multiple timeframes,” says Dermody. Clients can also apply AI capabilities to Snowflake’s datasets.

Data Partnerships

MSCI partners with over 80 data vendors covering market, liquidity, and fundamental data underpinning factors. Clients can access this data and also keep an eye on any bugs and issues. “We pull in the user holdings alongside user data sources such as client-specific fixed income pricing models,” says Dermody. High-quality sustainability data, including climate factors, is included in the factor model framework. 

Regular Compliance Functionality Remains

The system’s extensive enhancements have not led it to overlook the essentials. It continues to carry out the day-to-day tasks, such as flagging up anomalies and limits, which are most often user-defined limits relating to compliance and regulation such as maximum position sizes or tracking error constraints.

Upgrade Considerations, Cloud, and Code

Generative AI is embedded in the core package for new subscribers but it´s not included in the standard package for all subscribers, who may need to upgrade.

Costs vary with usage and depends on whether clients have an existing license with Snowflake. MSCI and Snowflake are agnostic to the choice of cloud provider, which could include Google, Microsoft Azure, or Amazon AWS. Flexible APIs from Snowflake and MSCI enable users to use the computer language of choice when further analyzing risk results. Python has fast become the language of choice for data analysis, while Java or C++ is also commonly used for production workflows.

MSCI has also created an interactive, Gen-AI-powered chat agent that can write and execute Python code in order to answer client questions.

Outlook

There is plenty more in the pipeline. Dermody continues to meet AI startups in the San Francisco Bay area to share new functionality and features that could further empower clients to do more investigations faster. “We will continue to add more use cases and more users to create a consistent source of truth throughout the enterprise. Our goal is to find that needle in a giant haystack of thousands of data points for our clients,” he sums up.

This article is featured in HedgeNordic’s “(Em)Powering Hedge Funds” publication.