Past Big Data Bowl Recaps

THE NFL’S INAUGURAL 2019 BIG DATA BOWL

Congratulations to the winners of the NFL’s inaugural Big Data Bowl. Finalists presented their findings to members of the NFL league office, team executives, industry-leading representatives and league sponsors at the NFL Combine. The two grand prize winners received four tickets to a 2019 regular season NFL game and a $1,000 NFLShop.com gift card.

2019 BIG DATA BOWL WINNERS

COLLEGE ENTRY

Matthew Reyers, Dani Chu, Lucas Wu, James Thomson, Simon Fraser University – Routes to Success

• The group modeled play success rate and expected points under various passing route combinations. Using a technique called model-based clustering, the group found several complementary pass route patterns that could consistently yield positive outcomes, even when accounting for defensive formation and behavior.
• Key Stat: Through effective pass route combinations, an offense could control roughly 70% of the field.

OPEN ENTRY

Nathan Sterken – RouteNet: a convolutional neural network for classifying routes

• Sterken treated receiver routes as an image recognition problem, using a neural network to categorize each route. Once grouped, these patterns were compared to win probability added (the change in the offensive team’s chance of winning the game before and after the play).
• Key Stat: The flat-in-post route, a staple of the Steve Spurrier days at the University of Florida, was the best three-receiver route combination.

2019 FINALISTS

COLLEGE ENTRY

Kyle Burris, Duke University – A trajectory planning algorithm for quantifying space ownership in professional football

• Burris used metrics like player speed, direction and acceleration to chart the space occupied by the 22 players on the field. One example highlighted a 64-yard touchdown pass from Derek Carr to Johnny Holton to show how Holton’s speed and direction indicated that he was moving towards an open space well before he looked open on the field.

• Key Stat: In the play above, Carr released the pass only 0.3 seconds after Holton beat his defender, suggesting the quarterback knew he was going to Holton before the receiver broke free.

Peter Wu, Brendon Gu, Carnegie Mellon University – DIRECT: A two-level system for defensive interference rooted in repeatability, enforceability, clarity, and transparency

• Wu and Gu merged statistical modeling techniques with potential changes in penalty calls and receiver catch probability to consider new standards for defensive pass interference and defensive holding.
• Key Stat: There appear to be two peaks in catch probabilities on pass interference calls, about 55% and 75%, which suggests the feasibility of a two-level foul system.

Jack Soslow, Jake Flancer, Eric Dong, Andrew Castle, University of Pennsylvania – Using autoencoded receiver routes to optimize yardage

• The group presented three unique ways to represent pass route data, including time series and shape-based clustering. Merging in-play and game-specific traits, the group suggested that hitch routes are generally an underused strategy for increasing efficiency.
• Key Stat: Roughly one in three Odell Beckham pass routes was classified as a 10-yard curl pattern.

OPEN ENTRY

Sameer Deshpande, Katherine Evans – Expected hypothetical completion probability

• Deshpande and Evans tracked receiver catch probability across entire pass routes. Their approach allows for an estimation of the receiver’s performance regardless of when and where the pass was thrown.
• Key Stat: On an 18-yard touchdown pass to Cooper Kupp during Week 1 of the 2017 season, Rams quarterback Jared Goff had another receiver that was more open than Kupp. Roughly 1.5 seconds into his drop-back, had Goff thrown to Robert Woods, the pass would have had a 92% catch probability.

Cathy Ha, Lucas Calestini – Efficient speed usage and the impact of fatigue in speed performance: an exploratory study

• Ha and Calestini looked at the link between play specific factors such as play type, game factors (home vs. away, game surface, weather), and player fatigue (rest since last play, intensity of last play) and their impact on player speed efficiency.
• Key Stat: Alvin Kamara had the highest speed efficiency of any ball carrier during the first 6 weeks of the 2017 season.

Adam Vonder Haar – Exploratory data analysis of passing plays using NFL tracking data

• Vonder Haar classified routes and defensive space allocation to identify which route combinations yielded the most open receivers. His approach using convex hulls to characterize defender spacing was particularly novel.
• Key Stat: The receiver generating the highest maximum separation was only targeted on about one in five plays in the first six weeks of the 2017 season.

Next Gen Stats data — which includes location, speed, acceleration, and velocity for all players on the field — was one of several data sets that are available to participants.

More than 2,000 data scientists from all over the world competed on Kaggle to predict rushing play outcomes during Weeks 13 through 17 of the 2019 regular season. Submissions to the 2020 contest were tracked using a leaderboard that updated weekly, with the participants who most accurately predicted how many yards rushers would gain winning the competition. 

2020 BIG DATA BOWL Winner

Data scientists Philipp Singer and Dmitry Gordeev from Austria captured the top prize of $50,000 with their highly technical approach in the NFL’s second annual Big Data Bowl.

Singer and Gordeev comprise The Zoo, a data science team based in Austria. Singer studied software engineering and economics in Austria, where he finished a Ph.D. in computer science. Gordeev studied at Moscow State University and graduated as a specialist in applied math/data mining. Both work at UNIQA Insurance Group in Vienna and have teamed up to win six gold medals on Kaggle.

2020 BIG DATA BOWL Finalists 

For the 2020 competition, the top submissions shared $75,000 in prizes, and top contestants presented their work to NFL club analytics staff at the NFL Combine in Indianapolis. Below is a summary of each presentation and a link to the complete entry.

GRAND FINALIST

Matt Ploenzke, Harvard

• Ploenzke used Next Gen Stats data to build interpretable model inputs based upon football-specific domain knowledge, ultimately highlighting the importance of ball carrier downfield acceleration and unblocked tackler distance and spacing. 
• Key stat: Among roughly 40 input variables, a ball carrier’s “effective acceleration” was the most important for estimating yards gained on a handoff play.

FINALISTS

Kellin Rumsey, Brandon DeFlon, University of New Mexico

• The battle between blocker and defender is often decided by leverage. In this paper, Rumsey and DeFlon define offensive and defensive leverage, and study the statistical properties of these metrics. 
• Key stat: In the first six weeks of the 2017 season, Blake Martinez (Green Bay Packers) was among the league’s best at generating defensive leverage. Martinez finished the season with the third-most solo tackles (96).

Graham Pash, Walker Powell, NC State

• Pash and Powell used kinematic data such as player positions and velocity to determine zones of control for both the offensive and defensive teams at the time of the handoff. These zones of control predict the probabilities of yards lost or gained and quantifies the risk involved with plays.
• Key stat: Robert Woods (Los Angeles Rams) and Raheem Mostert (San Francisco 49ers) outperformed the model predictions the most, averaging nearly three more yards than predicted over the 2017 and 2018 seasons.

Namrata Ray, Jugal Marfatia, Washington State University

• Ray and Marfatia measured the open space of the rusher at three time intervals — handoff, after a half-second, and after one second — to understand the association between open space and yards gained. Results indicated that the difference in the open space between the time of handoff and after a half-second or full second was a strong predictor of the number of yards gained. 
• Key Stat: Yards gained by the rusher increases by four yards on average for every one percent increase in the additional open area created within a half-second of the handoff.

Alex Stern, University of Virginia

• Using an advanced machine learning algorithm, Stern assessed the value of initial space created for the ball carrier by the offensive line. That space was then linked to linemen grades, and standardized by accounting for the number of defensive backs, linebackers and defensive linemen on the play, the defensive strength of the opposing team, and the running direction of the running back. 
• Key Stat: In 2018, New Orleans Saints center Max Unger received a top five grade according to Stern’s space grade rank for centers, despite being the 31st-graded run blocker by Pro Football Focus.

Caio Brighenti, Colgate University

• Brighenti computed each team's control of the field at the moment of the handoff to predict the outcome of rushes. Brighenti found that offensive control at the running back's expected point of intersection with the line of scrimmage was the most important predictor of run yardage.
• Key Stat: The critical factor separating successful and unsuccessful plays is ownership of the run gap at the line of scrimmage — even on plays gaining more than 10 yards, the difference in field control past the line of scrimmage was almost negligible.