clock menu more-arrow no yes

Filed under:

49ers hire Matt Ploenzke to their analytics team

New, comments

Ploenzke won the collegiate division of the 2020 Big Data Bowl

2008 NFL Draft Photo by Michael Zagaris/Getty Images

The San Francisco 49ers hired Matt Ploenzke to their analytics team, per ESPN’s Seth Walder. Ploenzke won the collegiate division of the Big Data Bowl. Winning the Big Data Bowl is a big deal as you have some of the smartest, most creative individuals around competing. The purpose of the Big Data Bowl is to continue the evolution in the use of advanced analytics. You can read Ploenzke’s full submission to the Big Data Bowl here.

He made several sample tables with in-depth explanations. Some of you may find his submission interesting. Here is a sample:

This report investigates relevant feature engineering in the space of predicting running play yards gained given NFL Next Gen Stats at time of handoff. In particular I focus on crafting features which generalize across all ball carriers (global features) as opposed to player-specific features, such as a Lamar Jackson effect. Inclusion of these latter features may make the trained model more prone to overfitting whereas the former feature types hold potential to generalize across all levels of football, teams, and styles of play. Indeed, it is through these generalizable features from which we as domain experts may interpret relevant effects and leverage model findings to “take it to the house” on offense or “bottle up the run” on defense.

Despite predictive performance lagging behind the top performing models submitted to the firstround competition, these hand-crafted features incorporate domain knowledge and tend to agree with football intuition. Undoubtedly improvements may be made to the model which could improve predictive performance however this often comes at the cost of model interpretability, thus providing a rather clear example of the interpretability-performance trade-off.

Why he won the Big Data Bowl:

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.

Ploenzke graduated from Harvard and will be on the Research and Development team for the 49ers.