Leveraging computational simulation for better stress testing

Leveraging computational simulation for better stress testing

By Justin Lyon, CEO, Simudyne

Justin, can you please tell the Risk Insights readers a little bit about yourself, your experiences and what your current professional focus is?

Following 9/11, I went to MIT to study computer simulation. I realised that I wanted to spend my life trying to deeply understand how important phenomena around us emerge and how complex adaptive systems work. I wasn’t convinced with the prevalent narratives, and with the explosion of data there had to be a better way for us to explore our environment and plan our best future.
I worked with institutions and companies helping them to understand how advanced analytics, simulation and artificial intelligence can help businesses.  Since then, I’ve incorporated this experience from companies around the world and dedicated myself to making Simudyne what it is today: the next generation simulation platform that is enterprise ready and used by global banks to help manage risk and be profitable. We cracked the engineering challenge of building the software that allows businesses to create high fidelity models of the real world, so they can test drive their decisions and see the likely outcomes in a safe virtual environment. In today’s world of complex interdependencies, I don’t think we can do business any other way.

What, for you, are the benefits of attending a conference like the ‘Stress Testing Europe Summit’ and what can attendees expect to learn from your session?

Stress testing and the regulation surrounding it is constantly evolving. I think it’s important to attend such gatherings and keep abreast of the latest trends but also challenges. Stress testing is a space where we believe simulation can make a huge impact, and we want to share that with you.

It’s been ten years since the financial crisis and regulatory requirement is only becoming more stringent. While banks have worked hard to get to grips with the requirement set by regulators, they have been slower to adopt new modelling paradigms.

I want to share with the audience the potential of simulation models for stress testing, and how simulation can offer new insights into a data set, without necessarily relying on historical data, and ultimately provide better financial resiliency and competitive advantage.

Can you explain the importance of computational simulations for better stress testing?

Risk management is concerned with ensuring a system is robust against a range of scenarios.  We can look to historical data and ensure that today’s system is robust to yesterday’s shocks.  But risk managers also need to be able to draw meaningful assumptions as to whether their system is robust to futureshocks.
To take into account the complexity of the market, we need to simulate it. We need to generate hypothetical states of the world and how this could affect a balance sheet, for example. We then need to run these scenarios hundreds of thousands of times to be able to gauge where the sensitivities lie and to better understand our risk. What if Brexit triggers a sell-off in the housing market? Or what if a geopolitical event in the Middle East causes oil prices to double?
It is these what-if questions that senior executives need to be able to ask in order to understand their environment, the tipping points and knife edges of risk. Ensuring that capital deployment is robust in a safe, ‘digital-twin’ environment is the best way to test outcomes before committing resource in the real world.

Can you provide a brief overview of some of the technology challenges you have come across and how these can be solved?

When we were initially working with companies on scalable simulation solutions, we realized these were resource intensive and costly.

This gave us the drive to develop the platform that is now Simudyne to bring an efficient, scalable solution to financial services that is easy to integrate with existing architecture.

On a mortgage model, for instance, you can create a model of the mortgage market to represent 10 thousand households but we’ve made it easier than ever to scale to all 27 million households, without re-writing your model, you just distribute it on a cluster of machines.

How do you think artificial intelligence can be combined with simulation models? What would be the key benefits?

We believe that simulation is the bedrock of AI. Together, you can extract more value out of your data and analytics than ever before.

Simulation’s key advantage over data-driven methods is that it allows us to forecast things that have never happened before and to run scenarios outside of historical bounds. Simulation and artificial intelligence are complementary for today’s business challenges: rather than rely on historical data and have our model fail when we need to predict future events, we can generate vast quantities of data using our simulation — even training the model in the extreme tails of the distribution — in order that our data-driven models can provide useful insight.

Just as pilots are ‘trained’ in flight simulators, so are machine learning models. We talk about ‘training’ them on data, just as the pilot’s brain is learning from the large amounts of visual and sensory information he is taking in while he’s on the flight simulator.

Our simulations engineer, John Hill, also wrote a blog post on this which can be found on the Simudyne website.

How do you see the risk landscape evolving over the next 6-12 months?

I see banks increasingly looking to generate business value from the risk infrastructure they have put in place since the financial crisis, especially within stress testing where banks invest huge sums of money.

I also see a shift in how banks will try to exploit their data, looking to move beyond the historical data experience to a new era of risk modelling to incorporate prescriptive analytics.

Leveraging this existing infrastructure to extract new insights and run multiple what-if experiments will allow risk managers to gain an unprecedented value from their data and to make radically better decisions. It’s not just a matter of complying with regulation, but also delivering competitive advantage.

You may also be interested in our Stress Testing Europe Summit…

Sign up for your free account: