Numera: The Innovative Data Warehouse Solution For Energy Trading

Aug 16, 2024

Discover Numera

BGK Analytics and ROITI have just revealed Numera, an innovative data warehouse solution for energy trading. The German energy trader EWE Trading GmbH was the first to implement it. In this interview with Numera’s architect Nick Perrott and ROITI’s CEO Ventsislav Topuzov, we will try to unveil the why and what of this exciting solution.

Meet the Experts Behind Numera

Nick Perrott

Born and bred New Zealander, with experience in mainframe operation and development, Nick Perrott is the architect behind Numera. He has experience developing software solutions for the retail, investment, central banking, and energy sectors. In 2004, he concentrated on development consulting within the Openlink Findur/Endur universe, emphasizing solid and dependable risk management systems.

Nick relocated to Germany in 2006 to learn the language and culture of his beloved. Lengthy and successful consulting roles focusing on Endur solutions with several large European Energy companies followed. He became fluent in German before returning to New Zealand in 2011 for part-time university studies and remote consulting. Nick’s CO2 footprint increased due to frequent business trips to Europe, but the COVID-19 pandemic and a few years of no business travel made remote work more popular.

Ventsislav Topuzov

Before founding ROITI, Ventsislav Topuzov was employed by OpenLink, the business that creates Endur, at their Berlin headquarters. His career in consulting within the energy sector started back in 2005 when he worked on a post-privatization restructuring project for a utility serving 1.1 mln customers in Bulgaria.

Through a project he was working on, he met Lyubomir Yordanov, who would become his future partner. Topuzov and his wife moved back to Bulgaria in 2012 when Yordanov suggested starting a consulting business together. One year later, they registered it.

Now, eleven years later, ROITI is a flourishing full-service IT consultancy working with some of Europe’s biggest energy trading firms.

How BGK Analytics and ROITI Joined Forces to Create Numera

Over the past decade, Nick has been interested in ROITI’s growth in the European energy consulting business area, having worked with one of its principals, Lyubo, at a client site. Since then, they have collaborated with ROITI consultants on various projects and shared late nights at various Brahaus in Germany. ROITI’s approach impressed Nick and he felt a strong affinity for their work. At some point in the past 3-4 years, Nick and ROITI started discussing the possibility of working together.

Identifying Market Needs: The Vision That Sparked the Numera Collaboration

The use of modern cloud-based data stacks and changes in the energy business have led to a potential business opportunity. This coincided with Nick’s flood of ideas on meeting the new requirements in the risk reporting and management space with the new technology. The complex process of grounding a new company in Bulgaria was thus started.

Numera addresses some Endur pain points but also offers solutions for meeting changing and increasing business demands, all in a very cost-effective way.

Revolutionizing P&L Calculations

The BAV P&L (Best AvaiLable P&L) approach used by Numera has been designed to provide the following:

  • P&L is always calculated using the “best available” source simulation results from Endur.  This approach means that Numera responds gracefully to any operational challenges by always providing the most accurate P&L numbers possible.
  • Automatic support for any partial or complete rerun (i.e. re-simulation) scenarios.
  • Full and automatic support for “start of day” and intraday P&L calculations.
  • Period metrics such as Dtd, Mtd, or Ytd are calculated in Numera rather than in Endur which lightens some of the operational load on Endur as you only need to simulate your “live” deals.

Enhancing Operational Efficiency with Numera’s Advanced Features

Numera ingests all Endur data in raw format, reducing the time required to incorporate new data into the data model.

Numera also supports high-volume data series from Endur, such as schedule data. This is vital for proper support of intraday/short-term trading areas, where volumes must be seen at the most fine-grained level (i.e., 15m or even 5m).

The Numera architecture enables data integration from multiple sources. In the core product, these can include clearing bank statements and other software such as CerQlar, which we will elaborate on later in this article. Additionally, integrations can be conducted as customer-specific extensions where necessary. This flexibility enables the team to support many new business cases, including reconciliations and scenarios where several systems are used for valuation or risk assessments.

What Sets Numera Apart from Competitors?

ROITI and Nick’s long track record of designing, building, and more importantly, supporting many Endur (and Endur adjacent) solutions provide a unique insight into the pain points of the ETRM business. This also brings a healthy respect for the operational risks that come with ETRM-focused applications.

This focus has been applied to the design of the Numera data warehouse solution so that it solves actual energy trading business problems rather than being a technology play. The unique Endur data pump technology allows the team to export all Endur data to Numera with an extremely small footprint in Endur.

The Technology Behind Numera Data Warehouse Solution

Numera Data Warehouse Solution Technology

The Power of Cloud Technologies 

Numera’s technology primarily focuses on Snowflake, a cloud-based analytical database that supports any volume of data. Snowflake allows for minimal storage costs, similar to the underlying BLOB storage costs for the cloud platform. It can be hosted on various platforms like AWS, Azure, or Google Cloud Compute, allowing customers to store their data within their accounts or tenants.

Snowflake processes raw data into numeric data, solving business problems through SQL transformations. It also offers Python code for extracting time series and volume data.

Snowflake is constantly developing, and adding AI functionality for training models and machine learning models. Once the team solves the “low-hanging fruit”, they might go into that, too. Snowflake also offers a Kubernetes cluster for containerized application code.

Customers can connect to the Numera database using various connection technologies, with Python and PowerBI being the main ones. With that being said, people can connect any language they like to the Numera database.

The team uses an infrastructure as a code product called Pulumi to set up and deploy the infrastructure for each customer, allowing them to deploy multiple stacks for different purposes. On the transformation side, the team uses an efficient tool called DBT (data-built tool).

Numera data pump is a JVS package, that runs inside the Endur system and pushes data to Numera, making it easier and reducing the impact on the system. Python is used for other data sources. Counting SQL, they have these three languages.

Numera’s Short to Mid-Term Roadmap

Short-Term: Numera Exports All Data From Endur

The short to mid-term roadmap for Numera focuses on exporting all data from Endur, including schedule data. This data can be granular, such as daily for commodity trades or instruments like gas, or power markets which can go down to 15 minutes. The current version of Numera exposes this data, allowing customers to use its time series generation function to extract the desired granularity or frequency.

The next step is to bring schedule data with detailed pricing throughout the day on the power product. This way we can calculate more detailed metrics, such as P&L by delivery period. Our goal is to calculate the P&L for 15-minute granularity. This has business benefits, especially for short-term power market companies building algo trading strategies and intraday and day-ahead markets. Numera also provides for lower latency P&L calculations, which is similar to what APM provides.

Mid-Term: Numera and Renewable Energy

In the mid-term, Numera is integrating all data, such as the new source system CerQlar. Why is this important? A GoO (Guarantee of Origin) is a digital certificate issued for power plants producing renewable energy. Energy suppliers can buy these certificates to label a certain amount of their energy as green.

CerQlar is the leading system for the GoO deal management for EWE Trading. For other processes such as P&L reports, the deals must also be booked in Endur. To achieve this, the GoO deals are forwarded from CerQlar UL (Universal Loader) to the CerQlar Interface running on the EWE AWS Environment which then forwards them to Endur via deal-booking-service and deal-interface. Numera plays a key role in unifying data. There’s no need to use different platforms anymore. Numera brings together data from Endur and CerQlar, making data handling easier, faster, and more convenient.

However, the speed of this integration is not yet up to speed, so it may be necessary to seek help from someone else. Overall, Numera’s short to mid-term road map aims to provide a more efficient and accurate solution for businesses in the power market.

What Next? Established Customer Base

The first goal is to grow market share and establish a wide customer base to integrate more data sources. Once this is achieved, there are many possibilities for providing advanced services to customers, such as using AI capabilities in the Snowflake platform. For example, building machine learning models for specific customer needs could be an interesting aspect of the company.

Another possibility is to provide alternative valuations for delta positions and daily valuations. Research has shown that training models that reproduce data from source data can provide real-time perspectives on P&L. Of course, scoping and risk factors must be considered.

The company’s philosophy is to focus on low-hanging fruit in risk departments and trading organizations, reducing pain points. This requires knowledgeable people who understand the business and the problems to be able to do data engineering. The company can do this, but there is still much work to be done before entering more speculative markets where they may not be experts.

 

Interviewee: Nick Perrott

Editor: Hristina Tankovska

(Visited 221 times)