top of page
  • Writer's pictureBruin Sports Analytics

Is the Onside Kick an Effective Strategy?

By: Leonardo Cardozo


Source: AP Photos

In the 2023 Super Bowl, the Eagles led by 10 points at halftime. After scoring on two of their three second half drives, averaging over a field goal a drive, they ended up losing by 3 points. Simply put, the Eagles could not stop the Kansas City Chiefs and Patrick Mahomes, as so many teams have failed to do since he entered the league in 2018.


Dominant offenses that can score a touchdown nearly every drive can almost always win the game in today’s NFL. With five years of evidence that defenses will not be able to stop Patrick Mahomes effectively, I propose an alternate solution: preventing him from getting the ball altogether.


In football, after a possession in which they score, teams have to kickoff to the other team, signaling the change of possession. But, if the kicking team kicks the ball farther than 10 yards and recovers it before the receiving team, they can retain possession. This is what is known as an onside kick, although it is hardly used in the modern NFL, since the tradeoff in field position is usually not worth it for a slim chance at recovery.


Is Onside Kicking Advantageous? When Should Teams Go Onside?


Overall, onside kicks are recovered at an approximately 17% rate (16.6% to be exact). However, onside kick recovery rate vastly differs depending on the context of the situation. In situations where teams are expected to kick onsides (classified by either being in an onside kick formation or trailing with less than 8 minutes remaining), only roughly 11% of onside kicks are recovered. In “surprise” situations (ones where neither team has a greater than 80% chance of winning the game), onside kicks are recovered an astonishing 42% of the time.


With such a high success rate, is it worthwhile for teams to go onside more often in neutral situations? At first glance, it seems like yes. On average, kicking teams gain 0.1 EPA (expected points added) when going onside in surprise situations. This makes intuitive sense. Since possessions are more important than field position, on average, teams gain 2.36 EPA from onside kicks and lose 1.61 EPA from unsuccessful onside kicks. Thus, in order for onside kicks to be a beneficial play, teams need to be able to convert them at a rate of at least 40.6%.



Unfortunately, this calculation means that onside kicking at a regular or predictable rate is not feasible for teams. Since recovery percentage is so low when teams know that an onside kick is coming, it is unlikely that a team would be able to successfully recover enough onside kicks if their opponents are preparing against it. But, in the current state of the NFL, when other teams do not know it is coming, it is ever so-slightly beneficial to kick onsides and get an extra leg up on the competition.


Which Kickers Are The Best At Onside Kicks?


Kickers try all sorts of tactics to help improve their teams chances of recovering an onside kick. Some try to kick the ball hard into the ground so it goes high up into the air, some try to add weird spin, and some try to recover it themselves.



By percentage, the best kicker at allowing his team to recover onside kicks in the past two decades has been Cody Parkey, who has managed to convert an absurd 60% of his onside kick attempts. Next best by percentage is Jeffrey Wilkins, and there are only five total kickers that have been able to eclipse that crucial benchmark of roughly 40%.



Another way of looking at onside kick success is by looking at kickers z-score, which factors in the number of onside kicks attempted to see how impressive the percentage actually is. While Parkey still remains the top overall spot, Neil Rackers’ ability to sustain a 43% recovery rate over 23 attempts stands out.



Results are similar for the worst onside kickers, with arguably the greatest kicker of all time Justin Tucker making the list. One thing is for sure: the ability to kick field goals does not have a direct translation to being effective at onside kicking. Just ask Bears fans.


Source: Washington Post

How Have NFL Rule Changes Affected Onside Kicks?



In 2018, the NFL made two significant rule changes to onside kicks. First, they required that kicking teams line up with five players on each side of the ball, and second, they mandated that no member of the kicking team, apart from the kicker, could line up more than 1 yard from the spot of the kickoff, removing running starts. These two changes have dramatically altered how successful onside kicks have been. Before 2018, onside kicks were recovered at an overall rate of almost 19%, with a neutral situation recovery rate of nearly 45%. Since then, onside kicks have only been recovered roughly 9% of the time, with neutral recovery rates only 21%. In effect, these rules changes have killed the onside kick.



Due to the infrequent rate of success with which teams use them, NFL teams have proposed making rule changes to replace the onside kick with the choice to go for a 4th and 20. While these proposals may make the game more exciting, the easiest way to do encourage more exciting play would be to revert back to pre-2018 rules.


Can We Predict When Teams Should Go Onside?


Obviously, teams are more likely to be successful attempting an onside kick in an unpredictable situation, or one where their win probability is more neutral. Using a logistic regression model, I was able to predict whether teams were likely to recover an onside kick roughly 84% of the time. Not bad! However, out of the times that the model did give teams a greater than 50% chance of recovering an onside kick, they only actually maintained possession 55% of the time. This suggests that no matter what, it is hard to predict when an onside kick will be successful.


Next Steps


Future work could include using tracking and formation data to see if certain types of onside kicks are more effective than others, and to see how kickers can improve at onside kicks.


Data Sources


All data from this article comes from nflfastR

The code for this article can be found on my GitHub here.


302 views

Comments


bottom of page