bp T&S - Re-Imagining Fundamental Analytics in bp Trading & Shipping
Team members:
David Maerz, SVP Trading Analytics & Insight
Robert Doubble, VP Trading Data Analytics
Carl Hale, VP Programme Management
Dan Parisian, VP Fundamentals Modelling & Infrastructure
For and on behalf of the Trading Analytics & Insight and I&E dTA organizations in bp Trading & Shipping
Country: United Kingdom
Organization: bp Trading & Shipping
T&S is the energy and commodity trading arm of bp and is one of the world’s leading energy, marketing, operations, and trading organizations. We buy, sell, and move energy across the globe to provide integrated solutions to over 12,000 customers in 140 countries. With upwards of 300 ships on the water at any given moment for bp, T&S moves around 240 million tonnes of oil, gas, and refined products every year.
Awards Categories:
- Most Impactful Transformation Story
Business Challenge:
Immediately following the appointment of Bernard Looney as the new bp CEO in 2020, the company announced an ambitious net zero low carbon agenda and the transition from an international oil company to an integrated energy company. Trading & Shipping (T&S), the energy and commodities trading division within bp, is a key enabler of this strategic intent. With over 12,000 customers worldwide, and a business spanning crude oil, refined products, natural gas, power, LNG, biofuels, and low carbon, T&S helps keep the planet’s energy moving.
Its commercial success is underpinned by a world-class analytics capability, comprising a global team of 160+ analysts in Europe, the Americas, and Asia. They deliver actionable insights, advanced pricing models, and valuations of complex structured deals that inform the deployment of risk by the commercial teams. Possessing strong business acumen, seasoned market knowledge, and deep technical know-how, they are the backbone of our ‘analytics edge’.
Historically many analysts have worked in vertically integrated silos, sourcing, cleaning, exploring data, building models, and producing outputs largely independently of one another. This frequently led to parochial, duplicative solutions that were sometimes frail and often highly manual. Opportunities to collaborate, share best practices, or to seek peer reviews were limited, and the development of modular re-usable solutions to common business problems was a rarity.
With bp T&S mandated to grow revenue by expanding into new countries, entering new markets, and scaling up existing business lines, demand for analytics will only increase. Successfully navigating the energy transition will require an agile, flexible analytics capability, one that our legacy working practices and Excel tooling cannot provide. Eliciting change required a disruptive paradigm shift in our ways of working.
Business Solution:
bp’s new strategic direction provided a powerful catalyst for a radical rethink of our analytics working practices and organizational design. In 2021 we re-organized analytics along technical discipline lines, embraced Agile, spun up four multidisciplinary Agile Squads, and agreed that Dataiku would be the cornerstone of our modern strategic analytics tooling. In addition, we created a specialist fundamentals modeling discipline, one that would spearhead our transformation activity.
Dataiku was a natural choice for an enterprise AI platform for the T&S analytics organization. With its concept of ‘Clickers’ and ‘Coders’, it was well matched to a population equipped with a broad set of technical skills and differing levels of proficiency, ranging from Excel novices to deep Python experts. Dataiku’s emphasis on the collaborative development of end-to-end model workflows also resonated powerfully with our goal of empowering cross-discipline squads to reimagine our next generation of predictive models.
In 2021, Agile squads in London, Singapore, Houston, and Chicago kicked off our analytics transformation journey. Informed by a small group of enthusiastic product owners, the squads set about reimagining complex, high-value, and business-critical Excel models in Dataiku. Favouring progress over perfection, our goal was to continuously accrue benefits by engineering intuitive model workflows that benefit from superior automation and increased robustness.
We now have a growing number of business-critical models in Dataiku, executing intraday as new market data arrives without human intervention. Linked Excel Workbooks have been replaced by simplified workflows comprising both visual recipes and bespoke Python code, organized in logical Flow Zone groupings that afford standardization through design modularity and bespoke, re-usable Plugins. What’s more, model outputs are disseminated to traders via highly interactive self-service dashboards.
As we transform, we routinely engage with Dataiku to share feedback, seek technical reviews of our design thinking, and learn how their customers are tackling similar problems.
Value Generated:
Now 18 months into our transformation journey, we have a growing number of business-critical models executing daily on Dataiku with a high degree of automation. Traders can interrogate model outputs using interactive dashboards and experiment with custom market scenarios that would be impracticable in Excel.
By embracing Agile and fine-tuning it to our business context, we have been able to continuously accrue benefits in double-quick time. Furthermore, by eliminating manual processes for loading and preparing data, utilizing job scheduling, and embracing superior automation, we free up analysts from clerical tasks to instead focus on highly dynamic energy and commodity markets.
Through a process of continuous learning, we have identified design patterns for common recurring tasks that are ripe for modularization, either in the form of reusable Dataiku plugins, or by creating bespoke Python libraries. By building out a suite of shared components, our transformation trajectory is accelerating, with new models deployed more quickly. As our momentum builds, so does our business impact across the trading floors, as we transform legacy models at pace.
Our work has received high praise from senior T&S leadership, citing its ‘game-changing’ nature, as well as recognition from Franziska Bell, SVP Digital Technology. In the case of low carbon analytics, starting from greenfield, we have built out an entire suite of analytics on Dataiku which has very quickly delivered material value.
A key enabler of our success is a strong partnership with the central IT team. The provision of a robust multi-tenanted platform with ~150+ users is key to building confidence in Dataiku and critical to embedding our new ways of working.
Arguably, our trailblazing analytics transformation is demonstrating to both T&S and the wider organization how new digital investments can advance bp’s commercial strategy.
Value Brought by Dataiku:
Over the course of our 18-month transformation journey, we have retired 140 Excel Workbooks, eliminating 500 spreadsheet tabs in the process. By replacing onerous clerical processes with superior automation, we have saved 174 analyst hours per year. Analysts now have more time to focus on high-value analytics. Models now run more quickly and more often, allowing us to quickly disseminate actionable insights to the commercial teams in response to market-moving events. Scenario analysis allows front-line traders to quickly understand how changes to model parameters impact the numerical output, helping them to build greater trade conviction and to deploy risk with increased confidence.
Agile working practices allow us to accrue benefits rateably, unlike a waterfall-based approach. Analysts reap the benefits of manual work being taken out of the system, while traders gain from having access to more powerful tools to understand markets. Duplicative, siloed model development processes have been superseded by collaborative, cross-discipline working practices, and a centralized repository of models and libraries of reusable components. Company knowledge is institutionalized, and key person risk is reduced.
With our oil, natural gas, power, and low carbon models now in a single central location we can seamlessly construct cross-commodity views and generate new commercial insights that were impossible while working in siloes. Powerful machine learning algorithms and Dataiku’s ability to handle large data sets provide the foundation for building our next generation of advanced predictive models, something inconceivable in Excel.
Our team is enthusiastic and energized by what can be delivered through our new ways of working and by embedding Dataiku at the heart of what we do. Empowered and encouraged, the team will continue to employ Dataiku in innovative and novel ways to underpin the future commercial success of bp T&S.