Krishnen Vytelingum recently joined Simudyne from J.P. Morgan where he was a quant within the Market Risk team. Prior to that, Krishnen performed a similar role at Sungard Systems, modeling Market and Credit Risk. These were roles for which he was well suited having spent five years in academia researching agent-based models for the financial and energy markets. With over 30 peer-reviewed publications and over 2,000 citations, Krishnen has a Master’s Degree from Imperial College London and a PhD in Artificial Intelligence from the University of Southampton.
Here he shares his experience of moving to Simudyne for the opportunity to work with our cutting-edge technology, exploring what advanced simulations can show us about how tomorrow could look so that we can better optimise today’s decisions.
What is your current role at Simudyne?
I’m a simulation engineer building agent-based models (ABM) of complex systems that demonstrate to the financial services industry how this type of simulation can bring an edge to their business.
How did you end up working on simulation modeling?
I completed my MSc at Imperial College London and then went on to the University of Southampton where I studied for a PhD in AI. I spent five years in academia researching agent-based models within the financial and energy markets, so it was natural that I ended up working in this space. On completing my PhD, I moved into finance at Sungard Systems where I was a quant in the Market and Credit Risk domain, subsequently moving to J.P. Morgan modeling credit risk for a couple of years before joining Simudyne.
What was it like moving from J.P.Morgan to Simudyne and how was the transition from a global bank to a small scale-up?
It was a unique opportunity for me to apply my research in the micro-structure and behaviours of financial markets and use my expertise in agent-based modeling to build the next generation of simulation tools. I have also been leveraging my experience as a market and credit risk quant to use ABM for XVA modeling.
Frankly, it feels very much like being at the top of the wave of building next generation credit risk models. Surprisingly, it took me less time than I thought to adjust to a smaller sized company culture.
On my first two days in the office, I built a working trading model that generates synthetic market prices. It was surprisingly easy to get to grips with the Simudyne platform using the online tutorials. Without a platform like Simudyne, it would take a quant several weeks to build a similar highly scalable agent-based model.
What kinds of executive level insights do you feel simulation can provide that would be interesting to a financial audience?
It is difficult to capture the dynamics of complex adaptive systems such as financial markets. An agent-based model paradigm helps us recreate this complexity so that we can understand emerging risk more completely. That means that we can draw insights such as the irrational behaviours that could lead to market instability, for instance, a flash crash or more sustained market turmoil.
Agent based simulation allows us to generate very granular synthetic data that scientists can use to gain strategic insights. This includes understanding the behaviours that lead to undesirable outcomes so that narratives can be built that avoid those negative outcomes. You can only really achieve this with bottom-up models that reveal underlying, low-level behaviours.
Could we prevent or significantly mitigate the next financial crisis using simulation?
Exceptionally complex agent-based models that capture all of the various risk factors, behaviours and other variables can give us insights into non-linear risk. We have the potential to identify the precursors of future crises. We can also understand the right regulations that need to be in place to mitigate problems before they become unmanageable. It effectively comes down to running a lot of scenarios through scalable simulations in order to formulate the best plan (or regulatory framework) possible.
We have been working with the Mitre Corporation to replicate their financial vulnerability model on the Simudyne platform. This model was developed for the US Treasury in response to the Global Credit Crisis with the aim of better understanding the causes and reactions, and potentially avoiding them in future.
What is agent-based modeling and how does it work?
Agent-based models capture the structure and behavior of complex adaptive systems, such as financial markets, down to each individual agent. These agents represent a real-world entity, for instance a customer, a household with a mortgage, a bank, an economy, and so on. They work by modeling micro-behaviours and interactions between different agents that lead to complex, macro-level outcomes. So by understanding the interactions of behaviours of individual agents within a system, we have far deeper insight into the causes of macro outcomes.
What opportunities can using this type of simulation bring to the financial services sector?
ABMs provide a virtual, digital world in which to experiment and develop optimal strategic behaviours. A good example is best execution; it’s difficult to understand the market impact of any given decision, for example an order execution, without having the ability to fully test that decision. An adequately calibrated market simulator would allow a trader to do just that and much more, such as testing against very elaborate stressed scenarios or replicating how upcoming market structure changes might impact trading.
What models and simulations are you currently working on?
I’m working on a trading model that captures the lifecycle of a trade, allowing us to benchmark the best execution of an order and assess its market impact. This potentially will give an edge to trading desks. I’m also working on an agent-based XVA model to understand the impact of behaviours of banks compared to traditional XVA models.
Why simulation?
Simulations show us what tomorrow could be, so that we can optimize our behavior today, making better decisions and mitigating potential risks.
And when you are not simulating?
I love travelling, particularly anywhere where I can get outdoors and hike. One of my favorite trips was to some of the more remote areas of the Hawaiian Islands. I’m originally from Mauritius and I love kayaking just about anywhere.