Ilyé! is a highly accurate football results predictions engine. It extracts performance patterns for every player and calculates how they will play, dependent on who they play with and against.
Predictions are made more responsive to real-life context by considering cumulative player history, competitions, grounds and managers.
Ilyé! users can generate predictions for any possible combinations of teams’ line-up, and can react to changes and unforeseen events.
Ilyé! is the most comprehensive football prediction engine that models a match in exactly the same way that professionals analyse football games, just with more computational power and data points.
Leveraging second-by-second player data, as well as a database of off-the-pitch events, Ilyé! presents users with an innovative online platform to analyse the likely impact of a given match strategy on the final result. The platform helps football enthusiasts and professionals to easily access the power of artificial intelligence and machine learning to navigate the complexity of football strategies.
The basic block is the player and the basic data set is the detailed player statistics on and off the pitch. For Ilyé!, a player’s performance is represented as steps in a sequence of matches, injuries and so on. Ilyé! represents a player as having some natural capability, that evolves over time, which is affected by teammates, tactics and the opposition.
Because Ilyé! is built as a sequence prediction engine, it automatically learns the effect of a player’s history: a longer playing carrier may translate in more experience, a longer playing stint in a single season may translate in fatigue and risk of injury. In effect, Ilyé! remembers what happened and how it mattered.
So for every situation, Ilyé! is capable of adjusting its expectation of player performance.
The next block is the team. Ilyé! can quantify the effects of interactions between players. A player does not necessarily perform consistently, regardless of where on the pitch they play. Therefore, a team balances its performance depending on its lineup and tactics (positioning of players and running pattern).
Ilyé! predicts team performance as the next step in a sequence of matches, subject to these interactions. And because a team is a collection of players, whose performances over time are also modelled, Ilyé! learns how a team performs depending on its line up, tactics and ‘run of form’.
Ilyé! represents a match as a contest between two teams within a context. The two teams are a collection of players who are influenced by their teammates and the opposition. A context is a collection of events and circumstances not directly related to a player’s performance, such as the pitch, the competition or ranking.
Ilyé! learns what is likely to happen in a match based on all of these interactions and their history. Ilyé! may ‘remember’ that a team tends to score more goals when Player A plays in a wide left attacking midfield position and is pitted against a right back over 1.85m tall and who played 50% more minutes than him over the previous four weeks. Such patterns are impossible to enumerate manually, so Ilyé! uses a proprietary deep learning technique to quantify them.