Why BI?

By Lee Campbell
This paper discusses the concepts of Business Intelligence in the context of both risk management and CubeLogic’s product offerings.

What is Business Intelligence?

Business intelligence (BI) is a general term for the discipline of gathering, storing, analyzing, and providing access to data to help enterprise users make better business decisions. In recent years many large software vendors
have launched applications in this area to provide the user community with a large range of BI tools. Vendors such as Microsoft, IBM and Oracle all provide sophisticated BI platforms. Of course, BI tools on their own only
really provide the platform for something useful, in the same way that a programming language or database provides a platform. These tools need to be applied effectively and intelligently to solve data problems in
organisations.

An organisation can spend a lot of time and money trying to extract business intelligence information from their databases. Some organizations use a small army of data professionals and perhaps a dozen different software
packages to produce simple reports. Also, if the report doesn’t have the proper information, its creators have to start again. So why does BI help to solve these problems?

In very general terms, business intelligence tries to answer the following types of questions:

• What were the total sales of a particular product last year?
• How does our profitability for the first quarter of this year compare to the same time period during the past five years?
• How much trading was done in a particular sector, and how has that changed over time?

The first important thing to understand is why BI is any different to traditional databases and software packages as a way of answering these questions. What is the added benefit? What does BI really mean and how does it differentiate itself in providing better solutions to organisations for their reporting and data management needs? To understand the answer to this question is firstly to understand the typical problems in an organisation. An organisation typically has many systems performing various functions. In trading organisations there are typically multiple front office trade capture systems, separate market data systems, potentially multiple risk, and accounting systems. In larger organisations the data is split over many sites, time zones and even continents. Valuations can potentially be computed in different locations at difference times of the day. Various types of user need access to the data for various purposes. Traders, for example need access to live positions and pricing whereas Risk Analysts need views of the data across complete portfolios, trading books, products or
counterparties. Back office users require different views again. The spread of this data and the ever changing requirements make timely and accurate data availability hugely problematic. With BI solutions organisations can
have immediate access to data when making decisions quickly through:

• One single version of the truth — enabling confident decisions by integrating all of the data; bringing together a simple, accurate version of business performance.
• Turning data into action — giving users access to the relevant stores of data throughout the organisation, and turning it into insights they can act upon.
• Integrated data giving greater insight — allowing users to identify trends, key risks or trading opportunities that generate business by creating personalized reports for better data analysis.
• Invaluable insights — helping drive business forward into the future with proactive views of exactly what’s going on at any one time.
  What If and Stress Testing – helping to perform various potential scenarios quickly and efficiently to provide stress testing, regulatory compliance and proactive risk managed future-proofed trading.

BI is extremely effective because it brings all the data together from different sources, and then allows for extremely flexible analysis. Most operational databases are designed to store your data, not to help you analyze
it. Also, performance is a key concern when aggregating huge amounts of data. In some cases the analysis might require large scale simulations or very granular slicing. In all cases an advanced data aggregation engine is needed to enable this to take place on the fly. BI offers this performance through OLAP cubes. OLAP cubes are multidimension data structures allowing typically queries to be run 1,000 times faster than a traditional database query. Moreover, with the recent introduction of in-memory analytics, even faster performance can be obtained, as this technique essentially stores the data from all source systems in memory rather than on disk in a data mart or data warehouse.

Cubes

The OLAP cube is an integral part of the BI landscape. A cube is essentially a multi-dimensional data structure, specifically designed for reporting and analysis. Cubes allow users to access their data, build complex reports and
perform ad-hoc analysis. Cubes are structured into measures (quantitative values) and dimensions (generally textual attributes). Here is where things get really exciting. Because the cube contains all of the data in a preaggregated form, it seems to know the answers in advance. For example, if a user asks for total positions by delivery month and commodity, those numbers are already available. If the user asks for total exposure by
counterparty and trading contract, those numbers and names are already available. Think of a cube as a specialized in memory database that knows the answers before you even ask the questions.

Users can ask any pertinent question and get an answer, usually in extremely high speed. For instance, the largest cube in the world is currently 1.4 terabytes and its average response time to any query is 1.2 seconds. Slicing and dicing extremely large amounts of data on the fly for ad-hoc reporting – that is the big advantage of a cube.

BI for Risk Management

The discipline of Risk management covers many areas, from analyzing a trading portfolio, looking at logistical or operational risks or viewing credit exposure to trading partners. In trading organisations both market risk and
credit risk require detailed analysis of the trading portfolio’s performance and current open positions. With large trading books this typically requires a lot of computation, aggregating many positions in various ways.
Furthermore, when considering what-if scenarios and use of simulations to predict a worse-case of a particular risk measure many millions, or sometimes billions of computations and aggregations are required. Organisations often deploy specialist software to perform these calculations. Typically these solutions have pre-determined reports and perform large-scale calculations in batch processes as part of an end-of-day process. This of course means a long period of waiting for calculations to be completed and a very rigid set of results. If inputs are wrong or requirements change it can often be hours, or days before new results can be reported.

Another key requirement in risk management is in the use of time analysis. Time analysis includes the ability to predict what might happen in the future, or compare trends with historical data. In credit risk there are many
time-based requirements, from looking at an exposure through time in a “walk forward” scenario, to viewing limit utilizations through time, or finally to viewing peak exposures in future months including future payables, physical deliveries and key dates for title transfer of cargoes for example. In market risk terms time analysis is used in VaR reporting, and again when considering historical prices to determine future market scenarios. A major benefit of BI is the ability to perform time intelligence on data contained within the cubes. Time dimensions can be used to perform advanced calculations and hence makes BI a particularly good fit for many risk management requirements.

As we have seen, BI, and specifically cubes, are designed to perform data aggregation quickly and on demand. Due to this step forward in technology, the need of pre-determination and batch-based calculations can be
replaced with high-performance, on-demand aggregation and holistic reporting. For Risk Management, it is clear that BI offers a perfect fit. It enables users to determine their risk analysis requirements on the fly and see the results immediately. This means a much more flexible and proactive approach to risk management can be achieved.

Analysis Tools and Visualization

Organisations need powerful tools to manage and analyze their data. Microsoft’s Excel spreadsheet product has long been a popular tool for all kinds of data manipulation, calculations and analysis. However, to many organisations Excel creates more problems than it solves. Data is often spread across multiple spreadsheets with many different versions. From an audit perspective it can become problematic and mission critical data is far from under control. However, users like the power and flexibility of Excel and seldom wish to trade it in for alternative, more controlled software. BI tools, and particularly Microsoft’s BI platform, offer an alternative to the spreadsheet dilemma. Cubes can be accessed directly from Excel, connecting to a centrally controlled server whilst providing the power and flexibility of pivot tables. The cube itself can maintain a single version of the truth, and handle all the security requirements, whilst the end user is able to fully utilize Excel’s vast data analysis, pivots and graphing features.

Figure 1 – Using Excel with to analyse data in an OLAP Cube

Another great benefit of BI is in its use of management dashboards. Key information can be published in graphical form for senior management, again directly linking to a centrally controlled data cube. Cubes exhibit Key
Performance Indications (KPIs) to highlight pertinent information, again in graphical form. Many companies offer advanced BI tools which are fully compatible with cube technology. These range from smart dashboard and
visualization tools such as Tableau and Power BI. BI Visualization is hugely powerful in providing insight into data in a way that quickly highlights trends, unusual patterns or red flags. This medium is far more digestible than analysing large amounts of table based data or reports. This is particularly beneficial when dealing with large amounts of data, for historical trends.

Figure 2 – An example visualization dashboard

Conclusion

The Business Intelligence vendor landscape is large and growing. The recent 2016 Gartner chart shows the main vendors and provides an in-depth analysis of the strengths and weaknesses of each. However, all these vendors
provide a base platform that is one part of the puzzle. It would then require a significant investment to design of a suitable data model, and then an integration project to bring meaningful data into that model before any
reporting, analysis or dashboard visualization can even begin. Centralized Reference data is also vital in feeding the data model and providing meaningful analysis. All this requires a significant investment of time and money. The unique benefit of CubeLogic products combines the power of Business Intelligence with a pre-built data model and an application suite to provide immediate business benefits to Energy, Commodities and Financial
Services companies.

Figure 3 – Gartner Magic Quadrant of BI

About CubeLogic

CubeLogic provides Business Intelligence products and services in risk management for the energy, commodity and financial markets. The founders of CubeLogic are all renowned experienced industry specialists who have an impressive track record of developing and implementing global risk IT solutions. In the current volatile market conditions, CubeLogic addresses the increasing demand for robust, cost effective Business Intelligence solutions for risk management.