Deep Learning Systems
We enable neural synergies.
Our deep learning algorithms make it possible to learn from portfolios across different banks - while protecting your data. Thereby we enable precise risk modeling even when data is scarce. The special architecture of our models is tailored to the characteristic distribution of the risk parameters. Using probabilistic modeling, our algorithms give insights into previously unknown risk characteristics of individual positions and portfolios. Besides more precise forecasts, this information allows risk managers to identify potential risks at an early stage.
Our algorithms are based on neural networks. The network architecture is specifically tailored to the underlying problem. The networks can process various inputs, from macroeconomic factors to sentiment indicators to payment flows.
Our models are trained in the cloud. We deploy our cloud environment to your existing cloud account at the touch of a button. We also offer an on-premises solution or you can simply use paraloq's cloud.
You can interact with our models through our APIs. To operate our models, we offer extensive packages for Python and R as well as an Excel implementation and a graphical interface (web app). The packages allow thorough analyses of your portfolio and come with a comprehensive framework to validate our models.
Our models provide estimates at the granularity you need. Risk parameters can be calculated at segment, group or customer level. Our networks are also capable of forecasting advanced distribution parameters such as expected shortfall, value at risk, and median, in addition to the expected value.