swissQuant just won a prestigious award for its novel CleaRisQ platform which systematically runs an FHS-VaR model for Central Counterparty clearing houses (CCPs) in real-time.

The solution was successfully integrated into the EuroCCP system framework to significantly optimize initial margin requirements while optimizing risk outlays for clearing partners. Initial Margin is collateral posted to a CCP by clearing members to protect against future risk exposures for their open positions. It is a key line of defence in a CCP’s financial safeguards package.

What are the main differences between SPAN and FHS-VaR models?

Both SPAN and FHS-VaR are market risk models used by CCPs for initial margin requirements. Prior to the acceptance of contracts for clearing, each clearing participant is typically required to post collateral to the CCP to cover the deep-in-the-tail risk incurred over short horizons, namely the settlement period or margin period of risk known as MPoR.

CME-SPAN is a versatile framework that has been around for 20 years. It exists in many flavours and can be adapted to many environments. The main characteristic is that it accounts for netting with calibrated correlations, where the correlations between risk factors and instruments are provided as inputs.

FHS-VaR conversely accounts for correlations in a data-driven way. Physicists would say it is closer to “first principles”.

The correlations are now an output and can be monitored, with metrics of higher relevance to the risks in consideration rather than using, say, the Pearson correlation which is inappropriate.

Netting between positions / portfolios of a clearing member can be allowed to a level in line with regulations and the CCP’s risk appetite. You basically solve the trilemma at the cost of data, which is anyhow at hand so that risk coverage and margin efficiency are improved with less calibration!

In addition, CCPs are asked to manage not only coverage but also procyclicality and the functionalities differ in each framework. swissQuant believes that a FHS-VaR model naturally allows more transparent and independent ways of controlling how conservative and procyclical a model is.

Then why isn’t the FHS-VaR model adopted faster?

Financial Market Infrastructure institutions (FMIs) usually act with a fair amount of conservatism. For example, risk managers are used to and comfortable with interpreting the SPAN parameter sets, the input correlations, and so forth.

But the main barrier is that an FHS-VaR framework is meant to be systematic and to operate with little intervention relying solely on data. So, more development work is needed, like initial parameter selection, regulatory approval, go-live, and other actions before the benefits can be realized. Once the developments are completed, an FHS-VaR framework needs less maintenance and provides more transparency, in addition to the previously mentioned smarter coverage and higher netting efficiency. Because the initial development effort is sizable, rarely budgeted and includes an operational / skillset challenge, swissQuant gladly offers training and consulting in this space.

How many institutions are currently using a FHS-VaR model that you know of?

Eurex has operated Prisma since 2016. We co-developed the model and provided the system to EuroCCP for equity and derivatives. The HKEX equity clearing house is soon going live with a model which swissQuant developed for them. That is about it for live models. CME has delayed the release of their own FHS-VaR version called SPAN 2.0, and there are other clearing houses attempting to develop their own version.

How easy is it to transition from SPAN to FHS-VaR, and what are the advantages to cooperating with a trusted partner or using a tested and proven solution?

Once budgeted, next steps to transition from SPAN to FHS-VaR involve:

• designing an FHS-VaR model and getting it approved by internal and external stakeholders including regulator and clearing members,
• adapting the current system needs,
• if using real-time, which is mandatory under ESMA, new operations and an adaptation of data workflows need to be addressed, and finally
• training of the different operation lines to change their deep understanding of the parameter sensitivities and the monitoring.

Overall, a challenging undertaking, but this is where swissQuant steps in and does the heavy lifting. With expertise and guidance, the transition is well worth the effort as volatility in the markets can be met with the most modern version of margin modelling available, resulting in more accurate risk coverage and margin efficiency.

Is CleaRisQ only FHS-VaR or are there other models too?

This is a great question because there is a lot of misunderstanding with FHS-VaR. And there is of course a lot more which the swissQuant CleaRisQ platform has to offer! We refer to the model in CleaRisQ as FHS-VaR because this is the main component for portfolio margining. We blend another component of slightly different flavour to include stress and temper procyclicality. But we also have add-ons for liquidity, concentration, and wrong-way risk.

The added value of swissQuant is that we know how to turn prices and a few more time series, into a robust and explainable measure of risk. This requires innovative proxying, outlier cleaning, volatility estimation, holding-period and scaling, like MPoR, aggregation, stress-period detection, and many others, to be robust and consistent with one another. In addition, CleaRisQ is a platform solution that satisfies real-time or intraday margin requirements, stress-testing within another risk framework, and collateral monitoring.

One of the most undersold achievements of the CleaRisQ platform is the real-time backtesting of the margin model: many people don’t realize you need 1’000-fold more computational power for this than for end-of-day calculations. We have engineered a backtesting engine that can run the production model, together with sensitivity analysis variants over all production portfolios plus hypothetical portfolios, indifferently, for 25+ years of market data and multivariate synthetic prices of arbitrary length.

The feedback from clients is that they are extremely pleased with the CleaRisQ platform, and especially impressed with the intuitive and surprisingly powerful backtesting interface!

“The replacement and integration of a risk system is a significant undertaking and launching with CleaRisQ is a major milestone for EuroCCP. SwissQuant have been a strong partner on this initiative, through the initial launch for cash equities and subsequent extension to derivatives.”

– Ed Hughes, Chief Technology Officer at EuroCCP

“We are excited to have extended our partnership with SQG. They have a strong reputation for their risk technology expertise and their CleaRisQ system enables us to execute on our European cash equity and equity derivatives growth strategy”

– Jonathan Tran, Chief Risk Officer at EuroCCP

If you would like to talk to one of our team about your company’s Risk Model Analytics needs get in touch with our team here.