BNG Analytics is a world-class big-data team for hire, combining more than 30 years experience in analytics, data engineering, and data science. Our work ranges from building highly scaled enterprise data solutions to consulting on specialized data problems. We’ve worked with high-growth startups to multi-billion dollar businesses, and with nearly every data type under the sun. We deliver creative, performant, reliable tools and insights.
With experience across a diverse set of industries, we are ready to bring unique and novel insights to your data-driven questions. Sometimes it's simply identifying how to ask the right question, or where to find the right data. Other times it involves building a heterogeneous model combining different approaches while maintaining consistency. Whatever your needs, our expertise and differentiated models will bring insights you can be confident making decisions on.
Need help designing scalable analytics infrastructure? We can advise and help you quickly deploy the suite of tools best suited to your specific analytics need, budget, and performance goals. We will assist you in analyzing expected workloads, implementing off-the-shelf solutions, avoiding performance bottlenecks, monitoring your servers, and ensuring availability. You'll be operating your own production-grade cluster in no time!
We offer full end-to-end data product design and build. With executive and product management experience to round out our technical expertise, we can help you design, scope, and execute on a robust data-rich application or analytics service to match your timeline and resource constraints. Our nimble team can get a prototype or beta running in a matter of weeks, or help revamp a legacy product to be more useful and performant.
Special focus on data, techniques, and applications in sports and sports entertainment. We build mechanistic models that predict player performance, forecast game and season results, and optimize in-game decision making.
Please contact us using contact form below.
Nested Dirichlet Model of Baseball Player Ability - Given a set of possible outcomes for any plate appearance, each player has a set of “true'' probabilities relating to these outcomes. These have nested Dirichlet joint prior and posterior distributions, which change over time via aging and noise. Player ability models are the basis for deriving play outcome, game, and season models and predictions.
Improved models for player ability and play results generate predictions of game results that are comparable to those derived from prediction markets and add information to the gambling line, overcoming particular biases of the market. This hierarchical model of uncertainty leads to uncertainty estimates with respect to season results that are much more accurate than those generated by point estimates.
NBA Player Movement Patterns and Similarity Scoring - Clustering and analysis of NBA in-game player tracking data yields distinct player position groups that map loosely to the five defined positions. This “positional fingerprinting” allows a more precise approach to player comparison across teams, games, seasons, and even the same player across different in-game lineups.
Brad has extensive experience designing and implementing state-of-the-art forecasting and optimization engines in fields including advertising, sports, marketing, finance, real estate and general operations as well as expertise with big data, in particular incorporating real-time decision analytics and behavioral modeling. The core of our Voodoo Sports suite was built out of Brad’s PhD research in Systems Engineering and Management Science on “Stochastic Modeling and Optimization in Baseball”.
Gaurav has spent his career building software platforms to enable data management, model building, and results analysis, with the ultimate focus on automating the easy parts in order to focus on the hard parts. His experience spans hydrodynamics, marketing mix, retail pricing, on-line advertising, spatial data products, automated vehicles, and NASA satellites.