All players are assigned a primary position for which they are expected to play the majority of the time.
The player categories are defined as: goalkeepers, full-backs, central defenders, defensive midfielders, central midfielders, attacking midfielders, wingers and forwards.
The data considered for each player is specific to the position type. For example, tackles and interceptions are considered to be key statistics for defensive midfielders, while successful dribbles and crosses are core stats for wingers.
The algorithm uses player stats from all matches, club and national team, and updates after these matches are played. During UEFA EURO 2016, the rankings will update during every match.
Weighting
As the Barometer acts as a form tracker, the more recent the match, the higher the weighting assigned to the data from it i.e. player stats from yesterday are given more significance than those from last week. If a player is not active (e.g. due to injury or non-selection), his ranking will decrease over that period of inactivity.
Additional weighting is attributed for quality of opposition using UEFA coefficients as the foundation of the values.
Once UEFA EURO 2016 starts, statistical events (e.g. goals, assists, tackles, saves) for the 552 players will be given a significantly higher weighting to maintain an accurate ranking of those performing well in France
And the above was what they used. The highlighted underscores the problem.
It's an attempt to show who is in form. If your last 2 friendlies were great, you shoot to the top.
Players who are rested because they are guaranteed spots in their squad fall farther behind. If you played in the Champs or Europa final, you get bumped up.
It's a who is on form rating and not a who is good ratings. That gives more credence to recency of performance than performance over a period.
So no, Milner is NOT statistically the 5 best player in the Euros and his performance over the last year statistically doesn't suggest he is.