• Systemic Risk Analysis for Regulatory Bodies:
      Systemic Risk Analysis will be the core application use case in the QualiMaster project. It is well understood that today’s financial markets are correlated to a degree significantly greater than historically and not simply within asset classes. Due to the Risk-On-Risk-Off nature of today’s trading, broad ranges of asset classes have become strongly correlated, so that diversification of risk is more challenging. As a result, new models of co-dependency going beyond the linear assumptions of correlation (e.g., copulas or multivariate point processes), are needed to properly explain cross asset behaviour, providing a more refined view of multivariate tail behaviour. We propose a market-based approach to real time systemic risk monitoring where key traded markets will be continuously analysed for cross-market correlation or co-dependency behaviour, as well as signs of market imbalance. Exposure models of institutions with systemic exposure will then be evaluated against market co-dependency behaviour to provide real time hot spot metrics. The analysis will be complemented with taking into account Social Web data for supporting and stabilizing the prediction of systemic risk, which we expect can be used for identifying additional indicators, and for contextualizing risk predictions. The idea here is to rely on the predictive power achieved and the early warning characteristics that have been proven for the Social Web as well as its ability to reflect societal developments. We envisage such a system as providing regulatory authorities with an early warning system for impending financial stress.
  • Risk assessment for institutional financial clients:
      The second use-case targets the need of institutional users like Hedge funds, Banks and Asset Managers. It is suited to cover both, pre-trading risk analysis and real-time, real money trading risk analysis. The produced system will be integrated with the existing analysis and trading application of SPRING. The added value will be the multi-variant and multi-market approach, which is expected to add important insights about systemic risks, and help avoiding fatal losses of capital under management. A collateral outcome is that this will help stabilizing the capital markets at the roots, i.e., within the trading system of the financial industry. In order to help users to discover, quantify, and implement the risk factors and rules into their trading models, QualiMaster will enable scalable multi-stream and multi-market analysis. The main data sources will include market data streams of the finest possible granularity (i.e., ticks), of up to level 3 quotes. On these streams, we will be continuously running a toolset of proven, yet computationally expensive algorithms for risk analysis, i.e., pattern-recognition algorithms, Bayesian network models, and Markov micro-simulation models. The volume of the data to be processed, in combination with the large number of markets that need to be monitored in order to mine latent risk factors, makes executing these algorithms challenging and monetary expensive, even considering the state-of-the-art machinery and cloud solutions. As such, state-of-the-art risk assessment solutions (e.g., R2 Financial: NxR2, PerTrac RiskPlus for Investors, and Finanalytica Cognity) are limited to a single market analysis, and to the portfolio of the user. They cannot provide a generic and multi-market systemic risk analysis, and cannot help on discovering and implementing new risk factors.

Exploiting the QualiMaster infrastructure, we will also consider novel financial approaches that have not previously been considered for multi-market analysis, such as the detection and exploitation of causal forces in the evaluation of extrapolation methods. Even though it has been shown in the past that causal forces can positively influence accuracy, this has not yet been tested for multi-market analysis, due to the increased.