Since our partnership began in 2018, Next Gen Stats and (AWS) have been at the tip of the spear when it comes to the applications of machine learning and football tracking data. The 2022 offseason was no different. Working side by side with data scientists from the , Next Gen Stats is excited to give you a sneak peek into two new metrics we are launching this season, inspired by submissions from each of the last two contests:
- Coverage Classification
- Expected Return Yards
A special shout out to organizers of the five-time event, and , who took the modeling process one step further this year by replicating the work of a Big Data Bowl submission that became the baseline coverage classification model.
Just like we've done in each of the last two seasons, here's a breakdown of these two new metrics for the 2022 撸先生AV season.
Coverage Classification
When a defensive coordinator designs a play call, the coverage he chooses to use is a foundational piece. Coverage identification has largely been relegated to arduous All-22 film charting ... until now! Enter the NGS coverage classification model.
Inspired by a Big Data Bowl submission from -- tested and validated by Bliss, and improved upon by the AWS Machine Learning Solutions Lab -- the NGS platform can now identify specific man and zone coverage types using nothing but raw tracking data and machine-learning architecture seconds after a play.
Armed with these coverage tags for each play, we are now able to dive significantly deeper into the strategies that shape the game. In particular, we can compound our understanding of the passing game through the lens of the defense, helping to answer such questions:
- Which teams run the most Cover 2?
- Who are the best man-cover corners in the league?
- How are defenses game-planning against specific teams?
- How has a quarterback performed against similar coverage schemes?
By leveraging our receiver route recognition model, we can even identify league-wide trends pertaining to which routes are most successful against specific coverage types.
Never before have coverage-type analytics been available in essentially real time. By relying solely on tracking data -- instead of a ball-centric broadcast view -- Next Gen Stats can power in-game analysis as the action happens.
Expected Return Yards on kickoffs and punts
Hall of Fame coach George Allen once said, "Football is one-third offense, one-third defense and one-third special teams." When it comes to most football commentary, special teams trails far behind the flashier plays from scrimmage. The minimization of this phase of the game extends into football analytics, which have a rich portfolio of research dealing with a coach's decision to send out the punting/field-goal unit, but lack the details dealing with the actual dynamics of those plays.
The 2022 Big Data Bowl sought to change that, and as a result, we now have a new tool which evaluates the return and coverage units on punts and kickoffs. Building off the same framework that gave us Expected Rushing Yards, we applied this modeling structure to the return game.
Given the different dynamics of punt and kickoff returns, we created separate models for each return type. We can highlight the flashy returns for touchdowns and evaluate which returners are the most consistent at creating yards with the resulting Expected Return Yards models. But beyond that, we are now equipped to dive deeper into other aspects of the return game:
- Should a kick returner have taken the ball out of the end zone?
- Is a punt returner too conservative or aggressive in signaling for a fair catch?
- Which gunners are most effective at limiting the space for a punt return?
These are all insights provided in this new model that can peel back the layers of the hidden battle of field position that is often taken for granted.
More NGS news and notes:
- Weekly Position Power Rankings powered by Next Gen Stats are coming to 撸先生AV.com. The weekly top-10 list will rank a different position group each week based on key advanced stats from the NGS toolbox.
- Next Gen Stats is again working with every 撸先生AV broadcast partner this season. Keep an eye out for NGS powered by AWS metrics as you take in live action.
- Every Sunday during the regular season and postseason, 撸先生AV Network's GameDay Live will feature advanced NGS metrics in the live scoreboard. The 撸先生AV RedZone channel will also feature live win probability and fourth-down decision analysis.
Follow the on Twitter.
