Creator: S. Rudd
Method: Markov chain
Window: One possession
Off-ball information: Defensive states tagged in event data

Markov Chains

Creator: N. Mackay, Method: Logistic regression and GAM, Window: One possession

Possession-Based Model

Creator: K. Singh, Method: Markov-line, Window: Next 5 actions

Expected Threat (xT)

Creator: KU Leuven DTAI, Method: Gradient-boosted trees, Window: Next 10 actions, Off-ball information: Possession history proxies

Valuing Actions by Estimating Probabilities (VAEP)

Creator: J. Fernandez, Window: Next goal, Off-ball information: Full tracking data

Expected Possession Value (EPV)

Creator: Stats Perform, Method: Gradient-boosted trees, Window: Next 10 seconds, Off-ball information: Possession history proxies

Possession Value (PV)

Creator: American Soccer Analysis, Method: Gradient-boosted trees, Window: Two possessions, Off-ball information: Possession history proxies 

Goals Added (g+)

Creator: StatsBomb, Method: Gradient-boosted trees, Window: Two possessions, Off-ball information: Broadcast freeze frame

On-Ball Value

It is common for skilled soccer teams to control the ball for an entire game but never score a goal. Possession models look to further quantify what kinds of possessions are actually valuable. A simple version is a location-based possession value, which estimates the probability of scoring based on where the ball is at that point in time. As the ball gets closer to the goal, that probability increases.

University of Michigan - Ann Arbor

Google

While big data is increasingly prevalent in sports, soccer presents unique analytical challenges due to its fluid, low-scoring, and continuous nature. Unlike sports with discrete, turn-based events (e.g., baseball), it is difficult to isolate and quantify a team's or player's contributions solely through traditional statistics like goals and assists. Consequently, soccer analytics relies on creative metrics, tracking data, and advanced statistical models to measure and evaluate performance.

Soccer Analytics

Goals in soccer can be greatly influenced by luck or random moments such as near misses, deflected shots, errors, or referees. Expected goals measures the probability that a shot will result in a goal. Factors include angle and distance to goal, current score, header or not, during a counter attack, etc. This statistic can be used to more accurately determine the performance of a team. If the score was 0-0 but the keeper's had 20 saves each, that score could have been 3-3 or more on another day.

Expected Goals

Data that is dethatched from players and it more indicative of a managers performance over seasons or a teams history.

League standings and results data

Data on a players athleticism such 40-yard dash time, vertical, etc. Also, their goals scored, assists, salary, etc. All this can be used to make player driven models to determine the performance of a team as a whole.

Player details & characteristics data

Gathering this data is where a lot of novel computer vision techniques come into play. Determining frame-by-frame basis of player from both teams, locations, ball locations, action details, tackles, passes, direction, speed is part of this effort.

Game event and tracking data

Basic game-level statistics in soccer analytics encompass high-level, match-by-match data. These include team-level or match-level metrics such as line-ups, stadium attendance, goals, cards, and substitutions, which may not be directly attributed to individual player tracking.

Learn Before

Related

Learn After