Quantitative Risk Infrastructure

Continuous-time
risk analytics that
adapt when markets break.

Stokhos Labs builds Neural Stochastic Differential Equation models that forecast the U.S. Treasury yield curve in real time, cutting forecast error by 64.3% during the COVID-19 crisis versus industry-standard models.

U.S. TREASURY YIELD CURVE · MARCH 2020 REPRESENTATIVE DATA
Neural SDE Forecast
Actual
68% CI
64.3%
MSE Reduction
vs. ARX-GARCH during COVID-19 crisis
Time Resolution
continuous-time SDE vs. discrete steps
5
Yield Maturities
3M · 2Y · 5Y · 10Y · 30Y U.S. Treasury
How It Works

Neural SDEs: a model
built for crisis conditions.

∂t
Continuous-Time Dynamics

Instead of daily or weekly updates, the model evolves through a differential equation that responds instantaneously to new information, with no lag during fast-moving crises.

gφ
Learned Uncertainty

The diffusion network learns state-dependent volatility directly from data. When the market enters a crisis regime, uncertainty estimates widen automatically, with no manual recalibration.

Crisis-Validated

Trained on 2015–2019 FRED data, stress-tested on COVID-19 March 2020. ARX-GARCH MSE: 1.943. Neural SDE MSE: 0.692. A 64.3% reduction — with zero post-crisis retraining.

Request Access

Ready to see the platform in action?

We work with quantitative hedge funds, fixed-income asset managers, and institutional risk teams. Reach out to discuss a pilot.