Four years ago The Economist wrote a story about how messy IT systems at financial service organizations were a likely factor in the financial crisis. Silo but deadly points to situations like the British bank that invested $100 million to calculate aggregate exposure – but could only do it once a day.
Sadly enough, much of what the publication wrote about then is true today.
Data management is still an issue: Financial institutions struggle to manage data in a way that allows them to use it quickly and with greater transparency. Patching and weaving disparate data sources together is not yielding the transformation needed, and ripping and replacing systems is not financially or operationally feasible.
A better approach is a holistic one that uses the existing data sources and applies a framework of data management, data stores and other smart techniques to break down silos without physically disassembling them. This is the best bet for control of market, credit and operational risk.
A holistic approach relies on layering the data – with attention to keeping it clean and manageable. There are four layers to this approach:
- At the bottom is the data layer. Leaving all the data at its source and federating the relevant sources at the lowest level of granularity is a cost-effective option. One approach for federating is the low-cost, scalable “data mill” that stages the relevant data into commodity, in-memory stores.
- Being able to re-use measures in different contexts is particularly helpful. For this to work, organizations need a calculation layer. For example, VaR, which would typically be calculated by the market risk engines and only be available for market risk purposes, could be made available for use in non-market risk contexts, e.g. in spotting and stopping rogue trading.
- The context sensitive layer allows users to assemble metrics provided by the Calculation Layer, and manipulate the analytical context to identify revenue opportunities or unearth previously hidden risks. End users are not reliant on IT assistance, so they can quickly respond to regulatory or management requests or carry out modeling and analytics that were previously inaccessible to them.
- The presentation layer allows the user community to view and present information in the most appropriate format, ranging from desktop to mobile device. Users can carry out advanced analytics in a highly intuitive, interactive environment, and produce reports that provide unparalleled confidence in the information and subsequent decisions.
Further benefits of the holistic approach
This method is particularly advantageous when it comes to cybersecurity. Fraudsters always look for an organization’s data blind spots and inefficiencies. For example by creating risk information at a different time than when payment instructions are executed, the risk positions are open to being manipulated to hide fraudulent behavior
By looking at information in different ways and across diverse units, organizations can:
- Keep an eye on trading positions. When a federated data approach is used, it is easier to look at trading patterns and anomalous activities. A complete understanding of a firm’s overall position can mitigate the risk of rogue events, market abuse and insider dealing, and limit the exposure to a particular commodity, derivative type or counterparty.
- Apply high-speed analytics to detect suspicious trading patterns in real time. High-speed analytics can’t work in real time if data is housed in disparate locations and must go through numerous steps to be cleaned and organized.
- Use sophisticated techniques to find suspicious patterns. Fraud network analysis, neural networks, and logistic regression techniques work best with complete data. These techniques help organizations find patterns that link rogue individuals to others doing similar things and to accounts listed under different names. By using an event stream processing engine, organizations can compare data from watch lists, business rules and scoring engines and compare it to the trading data pouring in.
There is a growing need to better harness data analytics. Banks have an enormous amount of data at their fingertips. With analytics, they can detect trends and create KPIs to proactively counter cyberthreats. To learn more about the cybersecurity challenges faced by those in the banking industry, Longitude Research conducted a survey (on behalf of SAS) of 250 banking executives. (The survey results, and information from in-depth expert interviews, are in the report Cyberrisk in Banking.)