Bite-Sized DemosUnderstanding VaR Calculation

0:00

Narrator Hello, and thanks for watching. This video provides a brief demo and general overview of how to get VaR in Molecule. If you have any questions about Molecule that are more specific to your organization, we're happy to show you more. Reach out to us for a custom demo on our website at molecule.io.

0:21

Alex Cerbone Hello. Today, we're going to cover how we get VaR in the Molecule application, some of the components that go into selecting our VaR, and how we can interpret some of these results. So, getting to the VaR in the app, I'm going to come over here to my Reports screen, and at the bottom of my Reports should be my VaR calculation. So for me, that's all the way down at the bottom here next to my extracts. It'll pull up a screen that looks something like this. So we run a Monte Carlo and a Delta Gamma VaR side by side. You'll see at the top, it'll tell you what our horizon period is. So in this case, it's a one-day and the confidence interval that we're using. So this is a 95% value at risk measure, so we're looking at the tail 5% to see the maximum risk that we see in that one day future coming forward. For componentization, we will typically componentize by your book or by your counterparties. Books, we find the most common, and we'll bring in this diversification effect as as well, which is telling us how much our risk is being mitigated by the diversification between those components.

1:36

Alex Cerbone So we view these components side by side, and we get a total measure out here. On the bottom, we will see that we have a 30-day backtest that is automatically generated for every single one of our VaR measures. When we look through this, we expect to see in the 95% level, a break maybe once within this 30-day period. It looks like we came very close on the first. We dipped down completely below on the 17th. So I can go in, I can investigate what happened in my portfolio in the market on that day to have such a significant drop in this. Of course, this is a demo portfolio, so hopefully for your actual practices, you're not taking as risky decisions as we're generating here in our demo scripts. But you'll always have a backtest to look at how reliable that measure is. This also maintains a history for every day that these have been alive. So typically we need a 30-day historical period to start calculating your covariance matrices that are required for generating the VaR calcs. But once you've spent 30 days on the app, you'll have a history begin to generate and I can roll back day over day.

2:55

Alex Cerbone So if I want to look at these in a longitudinal study, then it would be very simple for me to do so. Let's say that my Monte Carlo VaR is $37 million today. It was $27 million a month ago, a year ago. I can look at the relative risk that I'm taking, translating to P&L, or am I taking risks without the upside reward of doing so? We can start going into the details of that a little bit further. If I want more details on the actual calculation, then I can come in here, download the full diagnostics. It'll have the forward curves that were used, the volatilities, the covariance matrices, everything that's going into this calculation will come out of that diagnostics report.

3:41

Narrator Thanks again for watching our demo video. If you have any questions about how Molecule will work for your specific business, please reach out for a custom demo at molecule.io.