Evaluate the standards between the Top Five through ratings of transferred players on whoscored.com

I am a student of Grade 8. I studied python programming and statistics including standard deviation in the last semester .  I used “怕砍不留名” ’s post on” soccer.hupu.com” as reference and with my father’s instructions, I tried to compare the standards between the Top Five of Europe.

It has always been controversial on the high and low of the Top Five’s standard and there were no exact results. Some people evaluate them by UEFA Champion’s League and UEFA Europa League but this way has its limits. The teams which take part in UEFA Champion’s League and UEFA Europa League are actually quite high-standard relatively in their own Leagues. This way cannot cover the teams—in the middle and low parts of their leagues—which cannot take part in UEFA Champion’s League and UEFA Europa League, so it is not very appropriate to say that this way of comparing is fair enough.

whoscored.com（https://www.whoscored.com/）is a website which gives most soccer games in the world—including Top Five—‘s players ratings, rankings, all kinds of data and data analysis. Although the ratings of whoscored.com is a little controversial but it is fairly objective because of data analysis based on the same algorithm. In the comparison I take  La Liga as a standard and calculate the statistics of transferred players between La Liga and EPL, La Liga and Bundesliga, La liga and Serie A, La Liga and Ligue 1. I  assume that the player’s power remained the same before and after the transfer, then I find out the relationship between the player’s rating and the Leagues’ average rating(mean) , the standard deviation of the season. After this  I judge by the two different results of the two Leagues to achieve EPL, Bundesliga, Serie A and Ligue 1’s level relatively to La Liga.

Take David Silva who transferred from La Liga to  EPL in the season 2010~2011 as example. There are ratings of David Silva of the season 09/10 in La Liga and the season 10/11 in EPL. Firstly I calculate the mean and standard deviation of the ratings of La Liga ’s principle players (players with total appearances greater than the average number of appearances in La Liga, the followings are the same)in season 09/10 which was 6.88（Full score is 10, the followings are the same) and 0.33, then I calculate the mean and standard deviation of the ratings of EPL’s principle players  was 6.88 and 0.30. For next step, I subtract 6.88 (the mean of the ratings of La Liga ’s principle players in the season 09/10) from 7.34( David Silva’s ratings in La Liga in the season 09/10). I divide the difference by 0.33(the standard deviation of the ratings of La Liga ’s principle players in the season 09/10) and I get that David Silva is 1.40 standard deviation higher than the average of La Liga in the season 09/10. Then, I use the same way.I subtract 6.88 (the mean of the ratings of EPL’s principle players in the season10/11) from 7.15( David Silva’s ratings in EPL in the season 10/11). I divide the difference by 0.30(the standard deviation of the ratings of EPL ‘s principle players in the season 10/11) and I get that David Silva is 0.90 standard deviation higher than the average of EPL in the season 10/11. Standard deviation of La Liga (1.40) minus the standard deviation of EPL (0.90). If the difference is positive, than the level of EPL is higher than La Liga ’s. If the difference is negative, than the level of EPL is lower than La Liga ’s. Also take Silva as example, the difference (0.5) means that his level in EPL relatively  to other players was not so outstanding as when he was is La Liga. According to this we can know that the average level of players of EPL is higher than La Liga’ s.  This is sufficiently accurate once the sample of transferred players is big enough.

Specific process of enforcement:

Firstly, I have to achieve all the ratings of the Five Leagues players (players with total appearances greater than the average number of appearances of the League) in the seasons 09/10 to 16/17. However, whoscored.com uses the incapsula system, I cannot use python databases such as urllib, to achieve the webpage. Instead I use AutoHotKey to simulate keyboard operations and conserve every web page´s content into txt files. The AutoHotKey codes are as follows:

#Persistent

global fileName := "whoscored.txt"
global pages := 29

*Space::
     file := FileOpen(fileName, "w")
     loop, 30
     {
         Send ^a
         Sleep, 500  
         Send ^c
         Sleep, 500
         file.Write(clipboard)
         Send {Click}
         Sleep, 5000
     }
     file.Close
Return

Secondly, read these txt files with python and achieve all the players’ ratings with python regular expression, write the ratings into dozens of csv files.

After that, read the csv files with python and calculate each Leagues’ mean and standard deviation in each season.

Last but not least, with the ways mentioned above, analysis the statistics of transferred players of the Five Leagues during the seasons 09/10~16/17 by excel.  As I said, I make La Liga as standard, I analysis the statistics of players who transferred from La Liga to other Leagues and vice versa.

python codes, csv files and excel files can be seen in attachments.

Conclusion:

It can be seen from the two tables that La Liga has been obviously the champion throughout these years, Bundesliga comes second, EPL, Ligue 1 and Serie A are the third, forth and fifth.

The picture on the left: the 2 to 8 row shows the level disparity between La Liga and other four Leagues at the specific season (positive number, then the League’s level is higher than La Liga’s.  Negative numbers means on the contrary) In this way, we can know from the bottom row that the average level of the four Leagues are lower than the Liga’s at 0.12, 0.13, 0.20, 0.34 standard deviations each.

The  picture on the right: with La Liga as the standard (horizontal line) , if a league has score below the horizontal line, then the league’s level at that specific year is lower then the Liga´s level at that year. As for a league which has score above the horizontal line, in the contrary.

My statistic results has some discrepancy with the results on https://www.uefa.com/memberassociations/uefarankings/country/

This table is the results on this website:

La Liga is still on the top place but Serie A and Ligue 1 are quite different than mine. This may have happened because that there were only 2 transferred players between Ligue 1 and La Liga in the season 12/13, as same as 16/17 season between Bundesliga and La Liga. So to speak, the extreme values observed on the right picture (seasons 12/13, 16/17), cannot prove the real standards of Bundesliga and Ligue 1.