# On Global Average

In this post, I take a closer look at the global average. What is it and what does it actually mean for us?

#### Introduction

From my regular posts, one might have noticed that average scores are always higher for top 10K FPL teams than for random selections. Top teams have better, more diligent managers and, therefore, regularly outscore casuals. Now let’s look at the bigger picture.

#### Average Points Distribution

I ran a script for the overall league table after the last three match days and took snapshots of all the scores for all the 3.06 million managers in the game and here are the results.

Each graph contains two lines:

1. a grey one depicts interval average scores for each 5% of FPL teams (roughly 150K teams), i.e. for the following ranges: 1-150K, 150K-300K, 300K-450K, and so on;
2. a blue one depicts cumulative averages, i.e. average scores for the following ranges: 1-150k, 1-300K, 1-450K, and so on; the last one for 100% is the overall average.

Please note, that these are overall ranks on the horizontal axes, and not GW ranks. Click on a graph to make it larger.

##### GW12 – final results
• Official average score: 55
• Actual average score: 55.02

(clickable)

##### GW13 – intermediate results after the matches on Saturday
• Official average score: 26
• Actual average score: 26.76

(clickable)

##### GW13 – final results
• Official average score: 50
• Actual average score: 50.82

(clickable)

#### Analysis

What conclusions can we derive from these data?

1. Rounding. First thing that strikes my eye is that FPL doesn’t actually use the standard mathematical rounding to calculate the average. They rather use the floor function, i.e. they round  average points scored down to an integer number. Therefore, actual average score is always slighly higher than the one that appears on our monitors.
2. Negative slope. Both lines slope down in all the graphs. There is always a hike at the end of the grey lines which reflects good results of recently created teams.
3. Overall average normally corresponds to the average points scored by the teams in the middle of the overall table, i.e. to the average score of the 50%-55% interval. Average points scored for the top 1.5M teams are higher, for the other 1.5M teams they are lower.

I’ve analysed overall league tables after the league updates. It means that such a downward slope also reflects rank changes. Indeed, teams with good scores go up in rankings increasing average scores for intervals of higher ranked teams. Teams with bad scores go down in rankings decreasing average scores for lower intervals. But what are average scores for each interval before such movements in rankings?

Well, it’s harder to create such a graph as we can’t sort the overall league table by rank as of the start of the week. But to get an idea, here is such a graph for my sample of 20K randomly selected teams for GW13:

Both lines still have a negative slope, but it is not as steep. Average score for the top 5% of FPL teams as of the end of the GW is 63.3, while average score for the top 5% of teams as the they were at the start of the week is approximately 58.2 points.

#### Conclusion

Depending on your overall rank, average points scored by FPL managers around you will be different. The higher rank one achieves, the tougher competition he gets as there are better managers around who score more points on average. It leads to the situation that every experienced FPL manager is already familiar with: over the season, gaps in the overall league table get bigger, and, with every gameweek played, it will be tougher and tougher to catch up with those who are at the top now.

In this post, I’ve only taken a look at 3 match days. Can things change from week to week? I believe this situation is typical for most weeks. Theoretically though, there might be some exceptional weeks when casuals outscore top managers. As I’ve pointed out in my latest post on the results of GW13, 60% of the average points were scored by just 10 players with high ownership. Such players usually have even higher ownership among top fantasy managers and are often referred to as ‘template’ players. If most such players fail in one single week and the likes of RvP, Sturridge, and Baines recover from injuries and bring point hauls, I guess we could see a different picture.