Who can enter the Big Data Bowl?

It is open to the public, subject to the competition's rules for each edition. Entrants range from students to professional analysts and engineers. Entries are judged on insight, soundness of method, and clarity of communication rather than a single prediction score.

What data does the Big Data Bowl provide?

Player tracking data recording the positions of all players and the ball several times per second, with derived speed and direction, event markers such as the snap, play context, and player information. Exact tables vary by edition because each release is shaped around that year's theme.

Can Big Data Bowl data be used outside the competition?

The datasets are widely used for learning and research, but they come with specific usage terms set for each release. Read those terms before using the data for anything beyond the competition, particularly any commercial purpose. Terms can differ between editions, so check the specific release.

What Is the NFL Big Data Bowl?

The NFL Big Data Bowl is an annual analytics competition run by the league, hosted on Kaggle, in which participants analyse player tracking data from Next Gen Stats around a theme set for that year. It is one of the few public routes to working with NFL tracking data, which records every player's position many times per second.

What the competition is

The Big Data Bowl is organised by the NFL and hosted on Kaggle, the data science competition platform. Each year the league releases a dataset built from its Next Gen Stats player tracking system along with a theme, a particular aspect of the game it wants participants to study. Themes have varied from one edition to the next, each focusing attention on a different phase or type of play.

Participants submit analyses, typically as notebooks combining code, charts, and written explanation. It is open to the public, and entrants range from students to working analysts and engineers.

The competition is structured around finding new ways to measure or understand the game, rather than purely maximising a prediction score. Submissions are judged on the insight they offer, the soundness of the method, and how clearly the work is communicated. Finalists are selected from the entries and present their work.

That structure is part of why the competition carries weight in football analytics: it rewards the combination of skills that the actual work requires.

Why it matters for access to data

Full NFL player tracking data is not generally available to the public. The Big Data Bowl releases are among the most substantial public samples, which is why the datasets get used for learning and research well beyond the competition itself, subject to their terms.

Reading past entries

Past finalist entries are a good way to learn how others structured an analysis of this data: how they defined a measure, validated it, and explained it. Read them for method and communication rather than to copy a topic, since each edition's theme changes.

What tracking data contains

Positions over time. For every player on the field and for the ball, x and y coordinates recorded several times per second throughout each play. A single play produces many frames, and a single game produces a very large number of rows.

Motion measures. Speed, acceleration, direction of travel, and the direction a player is facing, derived from the positional readings.

Play context. Down, distance, field position, formation, and play outcome, joined from play level data so that tracking frames can be analysed in context.

Event markers. Frames tagged with events such as the snap, a pass release, a catch, or a tackle, which let you align plays with one another.

Player information. Identifiers, positions, and physical attributes to join tracking rows to players.

The exact tables and fields differ between editions, because each release is shaped around that year's theme. Always read the data description for the specific edition you are using.

Working with it in practice

Normalise play direction first, because teams move in both directions across a game and raw coordinates are not comparable until every play is oriented the same way. Then align plays on a key event such as the snap. Most early mistakes with tracking data trace back to skipping one of those two steps.

What people use it for

Learning spatial analysis. The data forces you to work with time series of coordinates, which is quite different from the tabular box score and play by play work most people start with.

New measures. Many public ideas about measuring things like separation, pursuit, pressure, and space originate in or are refined through competition entries, because tracking data makes them measurable for the first time.

Portfolios. A finished, well-communicated entry is a strong demonstration of football analytics skills, since it shows data handling, method, visualisation, and writing together.

Research and teaching. Released datasets are used in courses and research, under their usage terms, as a realistic example of high-frequency sports data.

What it is not is a source of live data or a complete historical record. The releases cover limited periods and are shaped by the theme, so any conclusion drawn from them has to respect that scope.

This page describes a data competition and is not betting advice.

Sample size inside a rich dataset

Tracking data produces an enormous number of rows, which can make samples feel large. The unit that usually matters is the play, the player, or the game, and the release may contain relatively few of those for any specific situation. A measure computed from millions of frames can still rest on a small number of independent plays, and uncertainty should be reported at that level rather than at the frame level.

Frequently asked questions

What is the NFL Big Data Bowl?: An annual analytics competition run by the NFL and hosted on Kaggle. Participants analyse player tracking data from the league's Next Gen Stats system around a theme set for that year, and submit notebooks combining code, charts, and written explanation.
Who can enter the Big Data Bowl?: It is open to the public, subject to the competition's rules for each edition. Entrants range from students to professional analysts and engineers. Entries are judged on insight, soundness of method, and clarity of communication rather than a single prediction score.
What data does the Big Data Bowl provide?: Player tracking data recording the positions of all players and the ball several times per second, with derived speed and direction, event markers such as the snap, play context, and player information. Exact tables vary by edition because each release is shaped around that year's theme.
Can Big Data Bowl data be used outside the competition?: The datasets are widely used for learning and research, but they come with specific usage terms set for each release. Read those terms before using the data for anything beyond the competition, particularly any commercial purpose. Terms can differ between editions, so check the specific release.