Client
PotentiaPRO is an advanced analytics company providing predictive outcomes and comparative effectiveness information for professional sports
Services
Data Architecture, Data Management, Data Analytics, Product Development
Industry
Sports, Entertainment
Tech
Microsoft SQL, SSAS, OLAP cube, .NET, Python, C#, Microsoft Excel, AWS Amazon, iOS

Challenge

Fantasy football, an online game where participants assemble virtual teams of real professional players to compete based on the latter’s statistical performance, is extremely popular among NFL amateurs. Back in 2015, fans had few to zero tools to keep track of NFL statistics, which is critical as the number of points earned by fantasy “owners” in their game depends directly on the real NFL players’ performance in the season. The client turned to our data science team for assistance with developing and launching a data analytics tool so that fantasy owners could have all the available NFL statistics at their fingertips. The client wanted to arm fans with an app enabling them to compare one player to another, or players from NFL history to current rookies to make accurate, data-driven decisions during the draft rounds.

We took on the following challenges:
Provide data cleansing by modifying an avalanche of dirty data
Orchestrate data across heterogeneous sources
Architect a database warehouse for terabytes of data
Design a proprietary similarity algorithm to compare players based on different criteria
Provide reasonable query execution time
Provide fast data updates
Provide end-users with easy-to-navigate relevant statistics baked in a pixel-perfect app

Solution

A predictive analytics model to predict football schemes during a game, depending on various input criteria and using data from sensors inserted into players' shoulder pads and hard-won insights from Todd Steussie, an ex-player and VP of PotentiaMetrics, who took on the Product Owner role on the project
ScoutSight’s proprietary similarity algorithm provides fans with access to player stats combined with a powerful algorithm to understand how this year's NFL Draft prospects compare to current NFL players
Custom Cascading filters to provide data cleansing by detecting and replacing gaps, duplicates, and irrelevant data. Since the format of NFL data sources varies from season to season, a critical part of the project was to filter all available datasets to unify data
ETL to fetch out all the data on players, their characteristics, scores, events, and games from multiple heterogeneous systems, and transform the data into a proper format for further querying and analytical purposes
OLAP cubes in SSAS to store data for quick querying and analyzing terabytes of data, enabling fast, consistent, and interactive access to it
AWS integration to establish complex workflows between distributed systems, and guarantee solid data delivery to thousands of fantasy football fans
API implementation to connect the data sources with the end-user iOS application
An easy-to-use iOS ScoutSight application as a reporting system for data analytics; a responsive and clean user-friendly design
Load testing against the IIS server with a huge number of streams, queues. In order to gauge the performance of the app, more than 1,000 users were involved
Predictive Analytics Model for PotentiaPRO
PotentiaPRO solution

Impact

Thousands of dedicated NFL fans were provided with an opportunity to easily navigate all previously inaccessible and unstructured data in a single place
With the data-powered iOS app that generates predictive indexes, fantasy sports fans can now filter and compare real players’ stats — hence, to foresee how NFL Draft prospects will perform as rookies
The easy access to data also helps fans gain a competitive advantage in building a winning roster team for fantasy football
The Scouting Notebook feature allows users to save players, comparisons and entire rosters to hit better scores during fantasy draft rounds.

Latest projects