Jesse Davis’ lab proves “boot it” can put you 10 actions from goal

Imagine the opening kickoff of a World Cup match and a player intentionally sends the ball down the pitch and out of bounds on the opponent’s end. Casual fans might assume it’s surrendering possession. But if you were Jesse Davis, you’d treat it as a setup. Davis is a professor of computer science at KU Leuven in Belgium and head of its Sports Analytics Lab, which has been pushing soccer toward a data-first renaissance since the lab was formally stood up in 2014.

The core finding, from a 2024 paper titled “Boot it,” is the kind that makes coaches double take because it attaches upside to what looks like surrender. Using a training dataset of more than 1.4 million passes and some 60,000 throw-ins, partly from the 2022 World Cup, Davis’s team modeled the tactic with tree ensemble models (a decision-tree mashup). Their conclusion: when the ball is in the middle third of the pitch, kicking it out of bounds on your opponent’s side can put you within 10 actions of a goal. In a sport with 1,500 or more actions per match and very little scoring, “within 10 actions” is not a small claim. It is a compounding edge.

Why would anyone choose an action that hands the opponent a throw-in? The lab’s argument is essentially about sequencing and recovery. By forcing the restart in a specific territory, you increase the odds of regaining the ball in an advantageous situation. That is the pivot from “possession is everything” to “possession is context.” Soccer is fast and messy, and most actions do not directly produce shots or goals. So rather than tracking only end results, the lab frames strategy around probabilities and the chain of events that lead toward scoring chances.

This is also where Davis’s approach becomes more than a single tactical tidbit. Soccer analytics is hard because the link between an action and a goal is indirect. Rios-Neto, the data recruitment lead for Belgium’s Royal Sporting Club Anderlecht, points to the broader impact: Davis’s lab has helped teams evaluate rosters, assess how efficient (or not) strategies are, and uncover hidden tactical patterns. Those patterns include everything from the value of certain restarts to how clubs might quantify other behaviors that coaches historically debated with incomplete information.

Take long shots. Davis notes that one trend over the last five to 10 years is that the number of long shots has dramatically increased. The data, he says, lets teams quantify the probabilities of those outcomes. In the lab’s work applying a Markov decision process, researchers modeled English Premier League behavior where some actions are under control while others are random, which matters because soccer movement is rarely linear. Presented in 2021 at the MIT Sloan Sports Analytics Conference, the results suggest Chelsea could gain 1.6 more goals per season by shooting from distance 20% more often. Again, the headline lesson is not “shoot more” as a slogan. It is the conversion of uncertain instinct into measurable decision rules.

Now, zoom out to the industry-level incentive question that boards and operators should care about: who controls the data, and who owns the insight? Davis occupies a niche where he shares research freely through open-source analytics tools, but the academic role also gives him room to tackle the unglamorous work that makes analytics scalable: standardizing in-game data. The reason standardization matters is straightforward. If event data and tracking data are inconsistent, every new season turns into a custom translation project. If they are standardized, clubs can parse game footage more reliably and build winning strategies faster.

On the pro side, the lab’s influence shows up in the hiring and workflow shift. Davis’s work has permeated European clubs, including Belgium’s Club Brugge KV, and it has reached national soccer organizations in the US and Belgium. Hugo Rios-Neto describes the work coming out of the lab as “genuinely useful,” and he notes that clubs apply it for multiple purposes. Van Haaren, now director of football intelligence at Club Brugge, describes the collaboration as translating the team’s football philosophy into measurable, data-driven outputs. When a club assesses a center-back, for example, it may tally how often the ball ended up in the pitch area closest to the opposing team’s goal by combining event data (actions on the ball) with tracking data (player movement). That is a pipeline from match actions to player evaluation, supporting development and scouting.

Still, there is a catch that keeps this story from turning into a tidy “analytics wins” narrative. The source notes that soccer somewhat lags behind many other pro sports in collecting the data analysts need. Teams annotate video and software captures tactics details, but the system details may make sense only to the most d... (the excerpt cuts off). Even with rising sophistication, the fundamental challenge remains: soccer’s data ecosystem is evolving, and the clubs that build the best workflows and standards will likely move faster than those still stitching together bespoke interpretations.

So what should executives and operators take from Davis’s “Boot it” proof point? It is not just about one clever play. It is about the organizational shift from debating tactics by feel to treating decisions as measurable hypotheses. When a lab can take a counterintuitive action, simulate it with millions of passes and tens of thousands of throw-ins, and produce a specific operational consequence (within 10 actions from goal), it changes how a club should think about training, recruitment, and game-plan governance. In a world where the margin is thin and scoring is scarce, turning “maybe” into probability is the real renaissance.