Financial Risk
Building a Robust Risk Model for Institutional Digital Asset Trading
Bruno GILLET · August 24, 2026 · 3 min read
When portfolio margining becomes a risk modelling problem.
On 19 August 2026, STS Digital launched cross-asset portfolio margining for eligible institutional clients, following completion of the Bermuda Monetary Authority's material change process.
The development is significant because portfolio margining changes the way risk is assessed.
Instead of calculating margin requirements independently for each position, the framework considers the risk of the portfolio as a whole. Positions and collateral are assessed together, allowing genuine risk offsets to be recognised while keeping exposures separate where they do not offset.
The principle sounds straightforward.
Implementing it conservatively is not.
As STS Digital itself puts it:
“The hard part of this product is not the trading, it is the risk model behind it.”
Why the risk model matters
Portfolio margining can make the use of collateral considerably more efficient.
But that efficiency depends on one fundamental assumption: the model must correctly identify which risks genuinely offset each other and which do not.
This becomes particularly important in digital asset markets, where volatility can be significant and correlations between assets may change rapidly under stressed market conditions.
A robust framework must therefore go beyond measuring positions individually.
It must consider how the portfolio behaves as a system, including scenarios in which positions and the collateral supporting them deteriorate simultaneously.
Getting this wrong can create an illusion of diversification and lead to insufficient collateral precisely when market conditions deteriorate.
CAPAnalysis' role
CAPAnalysis was engaged by STS Digital to conduct an independent review of its risk model and provide recommendations aimed at strengthening its robustness.
Our work focused on challenging the framework from a risk perspective: examining the methodology, testing its underlying assumptions and identifying areas where the model could be reinforced.
The objective was not simply to determine whether the calculations worked mathematically.
It was to assess whether the model provided a coherent, conservative and defensible representation of portfolio risk suitable for an institutional trading environment.
This distinction is fundamental.
A risk model can be technically sophisticated and still underestimate risk if its assumptions regarding volatility, correlation, collateral or market behaviour prove too optimistic under stress.
From risk modelling to institutional infrastructure
In May 2026, STS Digital was granted a Class F (Full) licence by the Bermuda Monetary Authority under the Digital Asset Business Act 2018, following a three-year progression through the regulator's Test, Modified and Full licensing pathway.
The Class F licence represents the highest level of authorisation available under the Bermuda framework. STS Digital subsequently completed the BMA material change process before launching its cross-asset portfolio margining offering in August 2026.
Independent challenge makes models stronger
Risk models inevitably rely on assumptions.
The purpose of independent review is therefore not simply to validate a model's calculations, but to challenge the assumptions on which those calculations depend.
What happens when volatility increases sharply?
What happens when historical correlations break down?
What happens when collateral and the position it secures fall simultaneously?
And does the model remain sufficiently conservative when those assumptions are stressed?
These questions are relevant far beyond digital assets.
They apply wherever financial institutions rely on quantitative models to determine margin requirements, collateral, leverage or exposure.
A robust risk model should not merely explain risk when markets behave as expected. It should remain credible when they do not.