With our broad variety of data-tracking sources, like player position and heart-rate data, ball position data and many skills tests and questionnaires data, we are able to develop powerful AI-based prediction models. With our models we aim for better quality of decisions and player development to the next level.
Read more
Use data to get to substantiated decisions.
Not only for your team, but for your individual athletes, one by one!
All based on data from previous session, with an output in load towards the individual.
Which of your players in is its peak performance stage?
No more worrying about putting to much load in practices on individual level.
Not only for your team, but for your individual athletes, one by one!
All based on data from previous session, with an output in load towards the individual.
Which of your players in is its peak performance stage?
No more worrying about putting to much load in practices on individual level.
The xSP is the expected succes rate for each pass in a match. This predictive data model was developed by Forward Football, based on 6.900 passes of 11 youth matches that we have tracked as part of The Perfect Choice research program. We keep the model learning, by adding new match data from the same age categorie.
The model is based on the 6 parameters that you can see in the model above. The xSP is a number between 0 and 1 and rates how likely it is that a pass is going to be successful (received by a teammate or not).
When adding new match data, for all passes we compare its xSP value (model) to the true Successful Pass (tSP). The tSP is 0 or 1, meaning the pass is received by a teammate yes (1) or no (0).
This comparison between the true Successful Passes (tSP) and the xSP benchmark provides an objective assessment in the form of player’s Pass Performance Ratio (PPR). Are you above 1 in this ratio? That’s good, you have played better than the xSP benchmark! If it is under 1 there is still room for improvement!
1) in green
Pass player 6 to 8
– 420/600 passes succesful (historical data)
– Expected succesfull pass (xSP) = 0,7
– ‘Easy pass’
2) in red
Pass player 6 to 9
– 275/550 passes succesful (historical data)
– xSP = 0,5 (50% chance the ball is received)
– ‘Difficult pass’
PPR:
True successfull passes (tSP) = 13 (match)
Expected succesfull passes (xSP) = 12,6 (model)
PPR = tSP / xSP
= 13 / 12,6
= 1,03
Objective benchmark to judge and compare the pass performance of (a) player(s). Per game, but more interesting, the development over time.
1) in green
Pass player 6 to 8
– 420/600 passes succesful (historical data)
– Expected succesfull pass (xSP) = 0,7
– ‘Easy pass’
2) in red
Pass player 6 to 9
– 275/550 passes succesful (historical data)
– xSP = 0,5 (50% chance the ball is received)
– ‘Difficult pass’
PPR:
True successfull passes (tSP) = 13 (match)
Expected succesfull passes (xSP) = 12,6 (model)
PPR = tSP / xSP
= 13 / 12,6
= 1,03
Objective benchmark to judge and compare the pass performance of (a) player(s). Per game, but more interesting, the development over time.
Currently we are modelling a player’s contribution to an attack. Each pass within the line of attack is given a rating, based on the start and end destination of the pass. For each position on the field, a weight is known about the probability of a dangerous situation if the ball is in that position (expected threat). We expect the model to be ready by the 1st of March 2022.
Currently we are modelling a player’s contribution to an attack. Each pass within the line of attack is given a rating, based on the start and end destination of the pass. For each position on the field, a weight is known about the probability of a dangerous situation if the ball is in that position (expected threat). We expect the model to be ready by the 1st of March 2022.
