Past Games
This segment evaluates the previous games in a quite simply way.
If Team A has won Team B three times, has drawn two times and lost 5 times,
this results in a win probability for A 3/(3+2+5) = 30%. But the evaluation
of only previous games doesn't make any sense, since even if a team has always lost,
it can win every new game.
Goal Average
It can be statistically assumed, that every team shoots 9 times at the opponent's goal.
Some teams meet more often than others. That is to say the probability to score is higher.
In our example team A played 10 times against Team B, scored 12 goals, but conceded only 21.
Thus Team A has a probability to score 12/10/9, that is to say 7% per shot against Team B.
Team B, in contrast, has a probability to score 21/10/9=12,3% per shot.
Now one could calculate binomial distribution and conditional probabilities of the single
chances to win.
FIFA-Points
FIFA-Points show current playing strength of a team.
Now we must convert them into the probability to score. There isn't any common
method for it. We use the following conversion formula (FIFApoints/1000)^2, it shows a
number of goals per game. However it's independent from the opponent.