Transforming Pro Sports with Cloud Computing and Data Analytics
In This Article
Across industries, organizations with structures in place to capture, analyze and act on data are better positioned to find new ways to drive value. Those who can unravel the complexities of where to place applications and data can build and deploy models faster and more efficiently, apply artificial intelligence (AI) and machine learning (ML) to data sets, and roll out new applications at scale -- all of which can drive organizational agility and success.
This is true in professional sports, too. As pro athletes around the world continue to compete for glory on the pitch, diamond, gridiron, ice and court, games are increasingly won and lost in the front office.
The concept of applying advanced metrics to sports -- distilling thousands of situational in-game data points into statistics that make it easier to objectively assess athlete performance -- has been around since the 1950s and '60s. Bill James championed sabermetrics in the ensuing decades, but it wasn't until Billy Beane's "moneyball" approach to constructing a baseball roster became publicized in 2003 that data analytics became increasingly visible and vital in pro sports.
Today's teams ingest and analyze data from dozens of sources, and their analysts are always searching for new models to generate insights that might confer an on-field advantage. For example, front offices with mature data analytics capabilities use proprietary models and algorithms to:
- Track internal player performance (e.g., "Is player X improving or getting worse over time?").
- Analyze the strengths, weaknesses and strategies of opposing teams and players.
- Develop sport-specific models to enhance and optimize player technique, which can enable on-the-fly and long-term adjustments.
- Monitor player health and well-being (e.g., heart-rate, oxygen utilization, nutrition, routine compliance).
- Identify up-and-coming talent faster.
- Compare player profiles to objectively assess trade needs and opportunities.
- Project future performance and explain anomalies in past performance.
- Recommend medical intervention or rehab sooner.
- Track individual performance via global identification across a player's journey from amateur to pro status in leagues around the world.
- Arrive at more informed and objective contractual valuations.
Establishing the foundations needed to reach these outcomes -- both from data maturity and cloud computing perspectives -- can be challenging for large pro sports organizations. This is particularly true given many teams' day-to-day reliance on homegrown infrastructures; the overwhelming number of tools and services available to handle workloads on-premises, off-premises and across multiple public clouds; and the sheer volume and varied quality of data captured in sports.
This article explores how World Wide Technology (WWT) helped one pro sports team make a new world happen by modernizing its cloud architecture and analytics platform, enabling analysts to generate actionable insights for players and coaches in a fraction of the time it used to take.
Over the course of many years, an established pro sports organization in the U.S. built its homegrown, on-premises data management and analytics system one piece at a time -- an approach common in the industry.
Many companies around the world provide analytics, tracking and video data for this sport, so it was no surprise the organization was ingesting data from nearly 20 sources dispersed globally, with a wide variance in data quality and type.
Once collected, player data was managed centrally in a data warehouse, where each data point across sources needed to be matched to a global player ID. Tagged data was then processed daily in batch-driven cycles before data scientists would run their analytics to generate insights for players and coaches.
The organization's legacy infrastructure was causing several problems:
- Data management: Maintaining data integrity and accuracy while aggregating, consolidating and tracking player data across incongruent sources was difficult.
- Reliability and lag: With daily batch cycles taking up to 20 hours to complete, data scientists were unable to consume or act on data in real time.
- Batch processing failures: If something went wrong during a batch cycle (e.g., certain data didn't load properly), the batch would need to be rerun, further delaying the availability of insights.
- Onboarding data sources: Adding new data sources to the legacy platform was cost prohibitive and labor intensive, taking anywhere from 40 to 80 hours to add a single new source.
Each of these challenges delayed the time it took to get actionable insight into the hands of data scientists, front office personnel and players to make more informed decisions on how to improve the team.
A lack of cloud expertise
Aware of its technology limitations, as well as the widely publicized benefits of cloud computing, the pro sports team initially attempted to build a new data management and analytics system on its own by migrating its workloads to a cloud service provider platform.
While the organization employed a number of data scientists and analysts, it quickly realized its staff did not have the cloud expertise required to migrate, optimize and manage its data in the cloud. The team also realized it could benefit from a complete refresh of its data analytics platform.
After considering the help of traditional consulting firms, the organization determined it needed a single partner who could combine technical expertise in cloud computing and data science with the ability to execute its vision for building a modern analytics platform.
So the pro sports team engaged WWT, whose Multicloud, Analytics & AI and Application Services teams routinely deliver the strategy and execution needed to make new worlds happen for customers around the globe.
Begin where you want to end
Before talking to customers about specific technology solutions, WWT's experts preach the importance of stepping back to understand the "why" before the "what." An organization's answers to "Why do you want to enhance your cloud and analytics capabilities?" and "What outcomes are you hoping to achieve?" set the stage for the ensuing digital transformation.
To determine a cloud smart migration strategy, our team of cloud application architects, software developers, cloud consultants and data analytics architects needed a firm grasp on the way data was being collected across the existing technologies of the organization's legacy system.
Stage 1: Discovery and analysis
WWT began by taking the time to interview key stakeholders within the organization who would be using the new platform. This included high-level front office personnel, analysts, coaches, scouts and players. With stakeholder help, our multi-disciplinary team of experts mapped out:
- Desired Outcomes: The specific outcomes and business value stakeholders hoped to achieve with a new cloud architecture.
- Current state: The complex interplay of existing architectures, technologies and systems in use, including relevant diagrams, workflows and processes.
- Pain points: The shortcomings and bottlenecks stakeholders experienced from the legacy approach.
- Necessities: Features stakeholders wanted to keep and those they could do without.
Stage 2: Solutioning
With a current-state assessment in hand, WWT began designing a technical solution to solve the many challenges of the legacy system. Our solution consisted of two main components:
- Cloud data management: Architecting and implementing a new system to manage the team's public cloud resources.
- Data analytics platform: Architecting and implementing a foundation for a modern data analytics platform that could scale over time.
At a high level, WWT replaced the team's legacy data management system with a new cloud architecture that consisted of a dynamic data-ingest platform running on a public cloud. We also created open-source application programming interfaces (APIs) so the team could ingest new types of data without needing to create an end-to-end pipeline from scratch every time.
We leveraged Infrastructure as Code (IaC) Terraform to manage the various resources needed to integrate with a newly created repository pipeline for their software, allowing the team to spin up and down server resources as needed.
Using best practices from other customer projects, we also helped the team move away from its batch-cycle processing approach (where a large collection of data points is processed all at once) to a stream processing approach that enables data to be processed continually, minimizing the amount of reprocessing that needs to occur). Our Application Services team accomplished this by:
- Developing bespoke software, which the organization has now taken over.
- Configuring incoming data as code, which removed the need to constantly write new code.
Thanks to WWT's data management solution, the team no longer needs its data scientists to gain the technical expertise required to manage workloads in the public cloud. Their data scientists and analysts are now free to focus on developing new models and applying analytics instead of managing hardware and software.
Other aspects of WWT's data management solution included:
- Alignment: Revamping the team's analytics processes and realigning organizational responsibilities to more efficiently leverage the capabilities of the new platform.
- Extensibility, scalability: Future-proofing the new solution by incorporating best practices for developing new models and applications on top of what's already there.
- Consulting: Advising the team how it might incorporate ML best practices in the future to further enhance data reliability and accuracy.
Broadly speaking, WWT reimagined how this pro sports team ingests, manages and analyzes its data. So what does that look like on the field and in the front office?
The organization can now rapidly ingest and access consumable player performance data in near real time, cutting a process that took nearly 24 hours down to 15 minutes. Instead of players and coaches waiting until the next day to receive feedback from a game, they can digest analyst findings as soon as the game is over, or even mid-game in some instances.
New data sources consumed faster
The team's new data analytics platform reduces the time it takes to onboard new data sources, cutting a six-to-eight-week window down to a four-to-eight hour window. This has saved the team a significant amount in onboarding costs and improved the time it takes for data to translate into on-field improvement or success.
Fewer processing failures
With the new data analytics platform, the data processing failure rate has fallen from more than 30 percent to less than 3 percent. The new platform's stream processing approach surfaces any data failures much earlier. Plus, we gave the team the tools needed to expand on the ability to discover failures even faster.
By trusting WWT to design and execute a highly extensible framework it can build on for the next decade, this professional sports organization is well positioned to optimize athlete and team performance in the increasingly competitive landscape of leveraging advanced metrics in pro sports.