I’ve played through 16 seasons of league soccer in FIFA 09 with teams ranging from America in the Mexican Primera: Clausura to SK Rapid Wein in the Austrian Bundesliga. At the end of each season, I’ve recorded data on Points, Goals Scored (GF), Goals Allowed (GA), the most goals scored by an individual, Assists, and what type of goal was scored. These are the graphs I’ve made to examine the data. Most of them are scatter graphs to see if there is correlation between stats.
In this first graph, I compare seven stats from each team. Two stats that bear explaining are goals inequality and assists inequality. Typically, a few players score most of the goals for the whole team while others score none. My measure of this was the average of how far away a player was from the average goals per player. If there were 45 goals scored, and 3 players, the average goals per player is 15. One player has 25 goals, one has 10, and one has 2. They are 10,5,and 13 goals away from the average. The average of those three numbers is the goal inequality: 9.3. I had to do it by hand since the data was in the game and couldn’t be exported. It sounds complicated, but it’s less complicated than a standard deviation. The assists inequality was done in the same way. Unironically, this stat of sharing the ball was more evenly distributed than the goals stat was. In other words, a wider variety of people contribute to a goal than actually score a goal.
The x axis value: Goal Differential per Game is how many goals you win by on average. A 3-1 victory is a GD of 2. A 2-0 loss is a GD of -2. The y axis value: Points/GD is how valuable each goal scored or allowed is to the game. If you are in a 1-1 game and score, that goal is important. You are much more likely to win the game and get 3 points. If you’re winning 3-0 and give up a goal, you’ll still probably win and get the 3 points. We can see that in this graph. On the left side, if you’re winning by less than a goal a game, scoring/allowing a goal adds/subtracts 2 points in the standings on average. On the right side, the teams winning by 2.5 goals per game are only benefiting by 1 point for each goal they score. It’s diminishing returns to scale. Many of the goals they score are “wasted” on blowout wins so they count for little.
This graph shows that having a go to scorer is a sign of a good team. Each goal the top scorer has corresponds to 0.6 points per game for the team.
When you compare the goals for the highest scorer with goals instead of general points, the correlation improves. Interestingly, a goal by the highest scorer correlates with 1.4 goals for the whole team. I believe this is because a lot of goals for the high scorer is just a sign of a good team.
Here we have the correlation between goals scored and points. If you compare the slope of this trend line with the slope of the previous graph, the goals from the highest scorer correlate with 47% more points than just anybody scoring.
Goal Differential correlates better with points because points are awarded based on the goal differential. It’s not if you scored a lot of goals, it’s if you scored more goals than you allowed.
Goals scored had a 70% correlation with points, but goals allowed only has a 26% correlation. I’ve allowed between 0.7 and 1.5 goals per game versus 1.8 to 3.4 goals scored per game. Goals scored has an 87% higher standard deviation than goals scored. It may be that the goals scored is better able to reflect the team’s strength because each data point is subject to less noise.
It turns out that assists are not well correlated with scoring. Assisted goals and nonassisted goals have the same slope with a lower correlation than the all goals graph we saw earlier. The last graph makes it clear that scoring assisted goals has no bearing on team success.
The more unequal the goal distribution is, the more goals a team scores. This has a pretty low correlation. Good teams have go to players who score. Worse teams have to get goals from whoever they can.
The percent of goals that are breakaways is somewhat correlated to goals. Similarly to before, the high scoring teams have a go to way of scoring, and that way is breakaways.
On the other hand, the percent of goals that are crosses has a negative correlation with goals. I think this is because the worse teams can’t pass right through a defense. They are forced to the sides and then they can try crossing it in.
To compare with my stats from playing FIFA, I put together data from the English Premier League’s 2014-2015 season. Here’s what it looked like:
The correlation of .6814 is almost the same as my FIFA correlation of .6623. However the slope is 2.27 vs 1.39. In the Premier League, there is an even bigger effect from the highest scorer. This may not be a good sample because the values are only between 0.1 and 0.8. Small differences in the highest scorer’s goals are magnified, perhaps distorting the relationship.
All goals scored add 0.9 points to a team. That’s about twice what it was for my teams in FIFA. I think this is because the FIFA teams scored an average of 2.7 goals/game while the Premier League average was 1.3 goals/game. As I said earlier, the more goals you score, the less each one is worth.
Goals allowed has a better correlation with points than my FIFA results did. Premier League: 0.6181, FIFA: 0.2597.
There aren’t any insights here that would help you play FIFA better, but it’s interesting to see the relationships between goals scored and points, or that there’s no relationship between assists and points.