Financial managers can have a tough job keeping up with and mitigating risks. Market swings and changing regulations make it hard to predict losses, adjust strategies quickly, and keep portfolios on track.
For Mathieu Tancrez, Market Risk Manager at Crédit Agricole CIB, artificial intelligence is key to overcoming these challenges by helping make sense of past crises and create realistic what-if scenarios based on current market conditions.
Explore how Tancrez combines financial risk management with AI, and get his step-by-step guide to tackling real-world challenges where quick, accurate decisions are key.
AI in Financial Risk Management
Traditional financial risk management methods can be cumbersome and prone to error because they heavily rely on historical data and do a poor job of predicting future economic events.
This is where artificial intelligence can help: AI can quickly sift through huge amounts of data, consider current market situations, catch trends that might slip past human analysts, and give more accurate predictions on key future economic events.
Machine learning, for example, can dig into data like stock prices, credit scores, and economic reports in order to find patterns and more accurately predict risks. This helps financial institutions better forecast market changes, predict the chances of loan defaults, and find potential financial threats more accurately than traditional methods. And because computers can rapidly process and analyze large amounts of data — like news articles, financial reports, and market commentary — they can accomplish what humans could never do alone.
These days, artificial intelligence can quickly summarize important details, point out potential risks, and create realistic scenarios based on what’s happening in the market right now. This means financial risk managers can make smarter decisions faster, stay ahead of potential downturns, and adapt to changes quickly — without getting bogged down in data overload.
This initiative to transition financial management into the golden age of AI is being led by industry experts like Mathieu Tancrez.
Tancrez’s Journey from Small-Town France to Financial Risk Management
Mathieu Tancrez hails from a small town in France. He studied financial mathematics in engineering school, and soon after graduating, he landed his first job in risk management at Deloitte in Luxembourg. In 2021, Tancrez moved to New York to work at Crédit Agricole CIB, seizing the chance to advance his career in a global financial hub.
It was around this time that artificial intelligence was becoming more well-known to the public, and Tancrez quickly developed a strong interest in generative AI — particularly the tools being developed by Microsoft, Google, Meta, and MistralAI. Since he was working primarily in market risk, he wondered how this new, transformative technology could benefit his field, asking himself, “What tasks in market risk could be automated, replaced, or enhanced by generative AI?”
Tancrez quickly identified the perfect opportunity for improvement, which led to the creation of StressGen.
How StressGen Helps Manage Unpredicted Risks
Large banks are required by law to run stress tests and ensure they have enough capital to handle tough financial situations. These tests entail simulating how the bank would perform under certain hypothetical situations (and even historical real-world financial crises) and then measuring its results — which then informs strategists what actions they should take to avoid big losses, how to stay within the risk appetite of the bank, and how much liquidity and capital they need.
Mathieu Tancrez’s StressGen helps financial institutions quickly determine the impacts of these scenarios (like Brexit or the next U.S. presidential election, for example). It works by generating scenarios — both historical and hypothetical — based on past data and current market conditions. It highlights and measures the impact of what Tancrez calls “market shocks,” such as sudden changes in interest rates or stock prices that could trigger a significant effect on a portfolio’s value.
“By simulating these shocks,” Tancrez explains, “StressGen allows financial risk managers to anticipate how their portfolios might react to unexpected events.”
“You could ask for five different scenarios, both real and hypothetical,” he elaborates. “The tool would then give you various details, such as a description of the scenario, the period and market data for the different risk factors, and any shocks at play.”
This way, banks and firms can meet their stress testing requirements without having to manually scour decades of historical information or coordinate research between various economists, risk managers, and data engineers.
In doing so, StressGen aims to take the heavy lifting out of stress testing, helping risk managers make more informed decisions, respect the risk appetite of the institution, and reserve capital for such crises.
Tancrez’s Step-by-Step Guide for Implementing AI Strategies in Risk Management
As Tancrez now knows, implementing AI into risk management isn’t as simple as plugging in a tool. It takes an informed, thoughtful approach in order to see results.
Define clear objectives
Before you start using AI, it’s crucial to know exactly what you want to achieve. Do you need AI to help make your risk predictions more accurate, speed up your decisions, or cut down costs? If your goals are unclear, you might end up with a tool that doesn’t really help. So take a moment to think about and discuss what success looks like for your team or business, and consider how AI can help you get there.
Start with Strong Data
It’s also important to understand that all financial risk management depends on strong data. AI is powerful because it can analyze heaps of this data to find patterns and offer insights — but if the data isn’t reliable, the results won’t be either. As Tancrez says, “The quality of the data is key: bad data leads to bad results.” To ensure reliable outcomes, make sure your data is accurate, complete, consistent, and up-to-date.
Examine Current Risk Management Strategies
In order to successfully implement AI, it’s crucial to understand your risk management workflow as it stands. This means figuring out what’s working well, where there are gaps, and what specific challenges need attention.
For example, if you struggle to accurately assess the risk of credit defaults among a large pool of borrowers, AI could help by analyzing detailed credit histories, economic conditions, and spending patterns to find early signs of potential defaults. This kind of analysis helps the firm take preventive measures before those risks materialize. “The goal is to identify where AI can really make a difference for your specific organization,” Tancrez says.
Select the Right AI Models and Tools
After you’ve set your goals and identified where artificial intelligence can make a difference, the next step is choosing the right AI models and tools. After all, it can be used for a wide range of tasks, from predicting future risks and classifying different types of data to clustering similar items and simplifying large datasets. It can also analyze large volumes of text to uncover hidden threats.
Since AI isn’t a one-size-fits-all solution, be sure to match the right tools with your specific needs. Communicate and brainstorm with data scientists and AI specialists to pick, customize, and deploy the models that will provide the most value for your risk management strategy.
Embrace the Unpredictable
In risk management specifically, Tancrez believes in expecting the unexpected. “Even a stress scenario that is extremely rare should be simulated and studied,” he says, “because it will save time to know how much is at stake and how to mitigate the losses if it ever happens.”
That’s why Tancrez used StressGen to simulate “Black Swan” events — those rare, unpredictable events like the 2008 financial crisis or COVID-19 — that traditional models might miss. As Tancrez puts it, “You don’t need to know how an engine works to drive a car, but you should always be prepared for an accident.”
Don’t Abandon Human Judgment
While AI is incredibly useful, it’s not a substitute for human judgment. Tancrez emphasizes that AI, especially in its current state, still has limitations. “Until artificial intelligence can fully train and improve itself,” he says, “human oversight is necessary to ensure that AI-driven decisions are sensible and don’t introduce new risks.” That’s why financial risk managers should use their experience and judgment to evaluate AI outputs and decide the best course of action for managing risks.
Continually Monitor and Refine AI Models
Even for routine tasks, financial risk managers and software engineers should regularly check their AI models to make sure they’re working well and not missing important data. For example, a stress test might overlook a new risk factor that just emerged from an evolving market (due to new interest rate benchmarks, for example). Regularly tweaking these models and keeping an eye on their results is key to making sure they stay reliable — especially as the market changes.
Leading Financial Risk Management into the AI Era
Artificial intelligence and machine learning are increasingly becoming key in helping financial risk managers find patterns, predict risks, and make smarter decisions in complex situations. Leading the industry in the development and integration of these tools are experts like Mathieu Tancrez. With the creation of StressGen, he’s paving the way for smarter, more efficient risk management.
To learn more about his blueprint for AI-driven success, follow Tancrez on Linkedin.
