EAGER: III: CIFRAM: Distributed Computing Approaches for the Analysis of Enterprise and Systemic Risk using a Financial Contract-Based Infrastructure

在最近的金融危机之后，监管反应和学术倡议刺激了对系统性风险测量的重要研究。然而，大多数系统风险模型受到可用数据和计算平台的约束。许多研究人员认为，如果市场参与者之间的关系和契约安排能够在微观层面上建模，那么理解系统性风险的可能性将大大提高。本研究的目的是通过建立底层分布式计算平台的模型（和实现），加速我们对利用颗粒数据（交易、位置和其他）的系统风险监测方法的理解。本研究的重点是实现一个参考架构和技术平台，旨在满足金融系统的特定数据结构，并解决与使用颗粒合约级数据分析系统风险相关的计算需求。研究工作集中在以下两个目标上：(1)使用可扩展数据库技术评估和实现颗粒契约级数据模型。这将需要分析可选择的设计模式和数据模型，评估数据库技术（包括关系模型，但特别关注NoSQL领域），以及使用最适合的技术实现数据模型；(2)基于底层数据库基础的分布式计算方法原型与评估。这将需要使用MapReduce或其他并行计算框架来利用许多处理资源的能力。ACTUS（算法合同类型统一标准）计算引擎将与计算环境集成，以提供颗粒事件向量和状态条件下的现金流。在此基础上，本研究将对汇总分析计算的各种聚合框架进行原型、测试和评估。欲了解更多信息，请参阅项目网站：http://www.usf.edu/business/departments/isds/projects/gsrisk

Following the recent financial crisis, regulatory responses and academic initiatives spurred significant research into the measurement of systemic risk. Most systemic risk models are bound however by the constraints of available data and computing platforms. Many researchers argue that the possibilities for understanding systemic risk will be greatly enhanced if the relationships and contractual arrangements between markets participants can be modeled at a granular level. The objective of this research is to accelerate our understanding of systemic risk monitoring approaches that leverage granular data (transactional, positional and other) by establishing a model for (and implementation of) the underlying distributed computing platform.This research focuses on the implementation of a reference architecture and technical platform designed to cater to the specific data fabric of the financial system and addressing the computational requirements associated with the analysis of systemic risk using granular contract-level data. Research efforts are focused around the following two objectives: (1) Evaluation and implementation of granular contract-level data models using scalable database technologies. This will entail the analysis of alternative design patterns and data models, the evaluation of database technologies (including the relational model but with specific focus on the NoSQL universe) and the implementation of data models using the best fit technology; (2) Prototyping and evaluation of distributed computing approaches based on the underlying database foundation. This will entail using MapReduce or other parallel computing frameworks to harness the power of many processing resources. The ACTUS (Algorithmic Contract Types Unified Standard) calculation engine will be integrated with the computing environment in order to provide granular event vectors and state-contingent cash flows. Building on the granular cash flow information, this research will prototype, test and evaluate various aggregation frameworks for the calculation of summary analytics.For further information see the project web site: http://www.usf.edu/business/departments/isds/projects/gsrisk