Using Machine Learning to Predict Current Ability

Current Ability

In our previous article Guide to Current Ability we showed how the current ability (CA) of a player is calculated using the value of their attributes and the appropriate attribute weighting.

We also released a handy Current Ability Calculator so that you could calculate the current ability of your squad.

The same calculations and attribute weights also formed the basis of our Role Ability Calculator.

The main limitation of the method developed was that it did not properly account for the scenario when a player can play in multiple positions.

Each attribute has a particular weight depending on the position of a player and its importance for that position. So tackling is more highly weighted for a defender than for a striker.

But what for a player who can play as both a defender and a striker? What is the attribute weight for tackling? It does not appear to be a simple average or the maximum weight.

Despite some extensive testing, we were unable to find an analytic method for calculating the CA for players with multiple positions.

However, since we published that article, Artificial Intelligence (AI) and Machine Learning (ML) have entered the public consciousness. Which led us to ask if we can use Machine Learning to calculate the current ability of players?