Latest News

Home Latest News EURO 2020’s Top Performers According to FSI’s Fantasy Scoring System!
#FSInsta
04-08-2021 by Vasilis Palaskas, Bill Mexias, Ioannis Ntzoufras

EURO 2020’s Top Performers According to FSI’s Fantasy Scoring System!

A Performance Evaluation of the EURO 2020 Soccer Players based on Fantasy Sports Points

V. Palaskas, B. Mexias & I. Ntzoufras

Fantasy Sports Interactive Ltd (Ltd) & AUEB Analytics Research Team, Athens University of Economics and Business

In Fantasy Sports, users select athletes from real-life sports events in order to create their own Fantasy Teams, which compete between them based on the actual performance of their athletes on the field. The performance of the athletes on the pitch is represented by an index (Performance Points) based on several actions performed during the match.

Fantasy Sports Interactive (FSI) is an awarded company providing users the opportunity for social interaction on sports events, as well as performance betting services in different formats, such as Over/Under (Performance Props) and Man to Man markets, for various sports such as soccer, basketball and American football, several national championships and European Tournaments like the Champions League and Euro 2020. The last couple of years FSI has developed a close collaboration with the Department of Statistics of the Athens University of Economics and Business (AUEB), the Department’s post-graduate programmes, as well as the AUEB Sports Analytics Research Team, through collaborative activities including internships and research papers, among others.

What FSI has done is to develop an in-house Performance Evaluation system where every athlete receives the corresponding performance points, according to the actions they perform in the game. These actions/skills are presented in Table 1.


Table 1: Performance Points System / Fantasy Scoring Rules by Fantasy Sports Interactive (FSI)


Statistical Analysis Methodology

The simple descriptive statistical indexes have a direct application to the sports sector, such as in the case of soccer players’ evaluation, which can be performed using statistical measures of position and dispersion (mean, variance, and standard deviation are some of the most well-known ones). Thus, sports fans and people involved in sports either professionally or on an amateur level, are able to discern and identify top performers, as well as under-performing players. Soccer players’ evaluation methods based on statistical techniques have appeared in the scientific publications of Sæbø, O. D., & Hvattum, L. M. in 2015, and Ian G. Mchale et. al. in 2012. Furthermore, the publication of Ian G. Mchale et. al. featured the modelling technique through which EA’s performance index was created (famous among many soccer fans and EA FIFA Sports video game players) which is also the official performance index of the English Premier League and has a similar sense with our system’s performance points. Lastly, Bayesian statistical modelling methods of performance points in Fantasy Sports have also been developed by Egidi, L., & Gabry, J. in their 2018 research. For brevity reasons, from now onwards whenever we refer to athletes’/players’ performance we shall mean their performance points.

In our analysis we follow a different research path which aims at the performance evaluation of the soccer players who participated in EURO 2020, using the performance points they received during the matches they played in, based on the evaluation system depicted in Table 1 as means to approach their overall performance. Essentially, the performance points that athletes receive according to Table 1 constitute a general index of their performance on the field, since this is a weighted index rating each skill (action) according to its importance and the position the athlete plays.


Our analysis is based on the following characteristics:

  • Athletes’ comparison will be implemented between athletes of the same position in order to compare them as fair as possible.
  • For each position on the pitch, we shall present their performance points in corresponding performance-consistency charts, where the horizontal axis represents their average performance, and the vertical axis represents a measure of their consistency (reverse quantity of standard deviation/variability).

To make this clearer to our readers, essentially, the higher the standard deviation (Footnote 1) , the higher the level of inconsistency of the player’s performance is, and vice versa, the higher the quantity of the reverse standard deviation, the higher the level of the player’s consistency is. Regarding the interpretation of consistency (vertical axis) we can use the average consistency of athletes playing the same position (which is depicted as a thin grey horizontal line in the following charts) as a reference point. Thus, the ratio of each player’s consistency to the average consistency of same-position players will provide us with the percentage change of each player’s consistency, compared to an ordinary player of average consistency (Footnote 2). In short, in all charts, athletes with a higher consistency score than the corresponding average of same-position athletes, present a more consistent performance than average.

Through the performance-consistency charts of athletes per position, we will be able to observe which athletes have the higher average performance among same-position players, but also to see whether this performance is consistent throughout the matches. Going forward we will present the best players per position and by extension we will put together the EURO 2020 Best XIs by the following ways:

  1. By selecting athletes whose average performance exceeds the corresponding average of same-position players, and choosing according to their average performance in descending order.
  2. By selecting athletes whose average performance exceeds the respective average of same-position players, but by choosing according to the ratios of their average performance to their variability in descending order (by the term “variability” we mean the standard deviation which is reverse to the consistency mentioned above).
Before we proceed to the presentation of the performance-consistency results of the soccer players and start building the XIs according to above criteria 1 and 2, we need to analytically explain some of our assumptions first.

Specifically, our analysis was based on:
a) Athletes whose National Teams proceeded to the Group Stage.
b) Athletes who played at least four matches as starters.
c) Athletes who played for at least 60 minutes in each of their matches.
d) Athletes’ statistics that were recorded until the end of extra time (in the cases there was extra time), but without counting their performance during the penalty shoot-outs.

The above criteria were established in order to have enough athlete’s data and subsequently a safe estimation and comparison of their performance. Specifically, we needed the soccer players’ comparison to be dependent on all the other skills/actions except for the rules concerning the number of minutes they play. Particularly, according to rules 1 and 2 from Table 1, if an athlete plays for less than 60 minutes, then they start off with one less point regardless of their remaining actual skill and actions they perform. For this reason, we used athletes having played for at least 60 minutes per match. We selected the rest of the criteria in a similar way.

Players Comparison According to Average Performance and Consistency per Position

In this specific section our goal is to identify the athletes achieving low or high performance in the Tournament’s matches respectively, and subsequently go a step further by also identifying how consistent the performance of high-performing players was.

For this reason, each chart per position is divided into four parts with the following interpretation:

A. Bottom left part: Athletes having a lower average performance and consistency than the respective average ones of same-position athletes, respectively.

B. Top left part: Athletes having a lower average performance but higher consistency than the respective average ones of same-position athletes, respectively (meaning they have a consistently low performance).

C. Bottom right part: Athletes having a higher average performance but lower consistency than the respective average ones of same-position athletes, respectively.

D. Top right part: Athletes having a higher average performance and higher consistency than the respective average ones of same-position athletes, respectively (meaning they have a consistently high performance).


Goalkeepers

In Figure 1, the most noteworthy characteristic is that Pickford, who played with England’s national team is the only goalkeeper whose performance as well as his consistency are higher than the average levels of performance and consistency. Such a result seems logical provided that England reached the Finals while managing to maintain a clean sheet in five out of its seven matches. However, another interesting finding is the average performance of Sweden’s Robin Olsen (also former goalkeeper of PAOK) which is the highest of all, while the performance consistency of Stekelenburg from the national team of the Netherlands, is also remarkable.


Figure 1: Depiction of average performance-consistency of EURO 2020 Goalkeepers In terms of performance points.



Defenders

In Figure 2 concerning Defenders’ performance, the most noteworthy characteristic is that the players with the highest average performance and consistency as well, were two players from Belgium (Alderweireld and Vertonghen), two players from England (Walker and Stones) and one player from Sweden (Augustinsson). For the latter, this finding shouldn’t be surprising, since while his team might not have been one of the top teams in the Tournament, his performance, however, was considered exceptional by several news media (MVP in Group Stage matches as a left-back!). On the other hand, Shaw’s and Alaba’s performances might not have been very consistent, however, their average performance was extremely high. Lastly, as we observe the two left parts of the Figure, we notice athletes whose performance was below average (bottom left part) with higher inconsistency, but also athletes with a consistently below average performance (top left part).


Figure 2: Depiction of average performance-consistency of EURO 2020 Defenders in terms of performance points.



Midfielders

About Midfielders (Figure 3) we notice that Sabitzer and Daniel James are the two players whose average performance and consistency are slightly higher than ordinary. Although the performance of Wales’s Daniel James is impressive, this fact is mostly due to his active participation on pitch, by winning many fouls and executing plenty of accurate crosses that had a positive impact on his performance index. Furthermore, the performances of Forserg, Mount, Pogba and Barela were also particularly high. The last two have the same average performance (Footnote 3)  but Pogba seems to have higher consistency in his one. Conversely, Sergio Busquets (Spain) has displayed a remarkably low performance, which makes sense, considering that he failed to execute a sufficient number of actions positively recorded in our scoring system, while at the same time he committed several fouls and received yellow cards, which had a negative impact on his performance.


Figure 3: Depiction of average performance-consistency of EURO 2020 Midfielders in terms of performance points.



Forwards

Lastly, among Forwards, Ronaldo’s average performance is remarkable (10,5!), despite the fact that Portugal didn’t fare so well in the Tournament, which is mostly thanks to the many goals he scored in various ways. Next to Ronaldo, regarding the average performance and the consistency levels, also stand Morata, Shick and Shaqiri. Regarding the last two, we notice that they have the same average performance throughout the Tournament, however, Patrick Shick’s one is slightly more consistent than Shaqiri’s one. Nevertheless, one of the most consistent players with a higher-than-average performance was England’s Raheem Sterling who significantly contributed to the number of goals and assists of his team achieved. The ones causing a rather negative impression are Chiesa and Griezmann regarding both their performance and consistency. Mbappe’s average performance was similarly disappointing, a fact that was rather unexpected before the tournament, considering his exceptional individual quality as well as France’s team overall quality.


Figure 4: Depiction of average performance-consistency of EURO 2020 Forwards in terms of performance points.



Best XI According to Average Performance per Position

According to Figures 1-4, the XI we get from the selection of players per position, based exclusively on the criterion of best average performance than the average of the respective position, is displayed on Table 2 and Figure 5, respectively. For most players in Table 2, the average performances were thoroughly commented on in the corresponding performance evaluation sections according to their position. Several XI positions in defence and centre are occupied by players of England’s national team, which was one of the Tournament’s finalists. This is mostly thanks to England maintaining a clean sheet for several of their matches. The absolute lack of Italy’s team players, who eventually won the Cup, is an especially interesting aspect. This is mostly due to the overall good team effort of Italy, which didn’t really depend on strong individual player performances. Lastly, it is worth mentioning that the XI was created in the scope of selecting players with the best possible performance among athletes playing the same position. For example, Spinazzola’s average performance was better than Maguire’s, but the left-back position was already covered by Shaw, therefore there was no room for this player in the XI.


Table 2: Suggested Euro 2020 Best XI based on athletes’ best average performance per position, according to FSI’s scoring system.

 


Chart 5: Best XI depiction, according to average performance per position for EURO 2020; average performance in performance points enclosed within parentheses.



Best XI According to Average Performance and Consistency per Position

The difference in the creation of this section’s XI is that the selection of players per position is made according to the criterion of the ratio of average performance to variability, given that all players available for team creation are players whose average performance is higher than the average performance of their respective position overall (see Figures 1-4). Specifically, when it comes to players whose average performance is higher than their position’s average (top and bottom right part of the figures), we select the ones also having a high ratio of average performance to variability. Meaning that the higher the athlete’s average performance, and the lower the variability is (reversely, the higher the consistency) the higher this ratio is as well. There is a possibility that this ratio is also high in cases where the numerator is low and the denominator is very low as well. Nevertheless, to avoid such confusions, we defined from the beginning as a minimum requirement that the individual athlete’s average performance exceeds the average exceeds the average performance of his position’s average one.

Observing Table 3 and Figure 6, the most significant conclusion is that for each field position at least one England player has been selected. Furthermore, Centre and Attack have two different soccer players out of total six, in comparison to the XI of Table 2. Additionally, we can see that in the previous XI (Table 2 and Figure 5), the Goalkeeper and Defenders may have the highest averages (which is a result of individual exceptional performances) but when it comes to the ratio of average performance to consistency, the respective players of the second XI are superior. Also, once again we do not see any players from Italy’s national team, for the reasons explained in the previous section.


Table 3: Suggested Euro 2020 Best XI according to the criterion of best performance in terms of ratio of average performance and variability per position, based on FSI’s scoring system (Footnote 4).




Chart 6: Best XI depiction, according to average performance and consistency per position for EURO 2020; average performance in performance points enclosed within parentheses.


Comparison to UEFA’s Best XI

In this last section, we put together UEFA’s recommended XI according to their criteria, and the players’ average performance (Footnote 5)  according to our performance evaluation system. In the comparison of UEFA’s XI with our respective teams, we observe the following:

  • Trophy-winners Italy have at least one player in each line of the UEFA XI.
  • Between our two teams and UEFA’s XI, there are 3 players we both recommend, (1 and 2 players in common with our 1st and 2nd XI, respectively). In our first team, our common player is Maguire, while in the second, we have Walker and Sterling in common - all of which are England players.
  • In Table 4, regarding our non-mutual player selections, their average performances as well as the combination with their respective consistency (see Figures 1-4) led us to - not - select them, based on the corresponding criteria 1 and 2 for team creation.

Generally, if one attempts to visit all the sports blogs suggesting the best XIs of the Tournament, they will notice that there are very few cases of commonly selected players, a fact happening due to the use of different statistical criteria, or even objective ones. Usually, the final UEFA “Best XIs” recommended by fans or media, are based on athletes’ best performance, or their performance in the finals (because of the importance of the match) and not on the overall performance profile of each athlete. For example, a goal in the Final is enough to send a player in the Best XI, even if he had a below par performance in the previous games. Based on the performance points used by FSI as an approach to the athletes’ overall performance, UEFA’s recommended XI has a significantly lower total average performance (48,60) than both our recommended teams have (73,48 and 60,90 for our 1st and 2nd XI, respectively).


Table 4: UEFA’s recommended Best XI for EURO 2020


Chart 7: Best XI depiction, according to UEFA per position for EURO 2020; average performance in performance points enclosed in parentheses.


Lastly, regarding which XI is considered to be the best out of the 2 we have recommended, we consider the answer to lie with the criteria that the reader deems as most important. Considering that EURO has a short duration, we consider as better the criterion 2, which is based on the selection of players displaying an above than average performance, but also consistency, to be “safer”, in order to avoid selecting players with truly remarkable, yet rare, performances.


Footnotes:
 1) Standard deviation is an index of dispersion, meaning it measures the distance of values from the mean (average).
 2) The percentage change actually results if we subtract one unit from the consistency ratio and multiply by 100. Thus, a player’s consistency ratio of 1.43 means that this player is more consistent than the average by 43%, while a 0.75 index means that he is less consistent from the average by 25%.
 3) In the performance-consistency charts of performance points per position, it is possible that some names are slightly shifted to avoid covering another name. These specific cases feature a small arrow indicating their actual placement.
 4) Contrary to Table 2, in Table 3 we also added variability values as well as the respective performance-variability ratios, as the XI creation is also based on these values in addition to average performance, as per criterion no.2.
 5) In Table 4, we have only included average performance and not those players’ variability, for a direct comparison with both of our suggested XIs to be possible.


Bibliography for studious sports fans:
  • Sæbø, O. D., & Hvattum, L. M. (2015, October). Evaluating the efficiency of the association football transfer market using regression-based player ratings. In Norsk IKT-konferanse for forskning og utdanning. 
  • McHale, I. G., Scarf, P. A., & Folker, D. E. (2012). On the development of a soccer player performance rating system for the English Premier League. Interfaces, 42(4), 339-351. 
  • Egidi, L., & Gabry, J. (2018). Bayesian hierarchical models for predicting individual performance in soccer. Journal of Quantitative Analysis in Sports, 14(3), 143-157.


About the Authors:


Vasileios Palaskas works as a Statistician and Sports Data Analyst at  FSI - Fantasy Sports Interactive, Ltd. He holds a MSc in Statistics from Athens University of Economics and Business (AUEB) and a Bachelor’s in Mathematics from the National and Kapodistrian University of Athens. He is currently working on Bayesian Modelling for several Fantasy sports, such as soccer, basketball and American football. He is also a member of the AUEB Sports Analytics Group and has recently published a scientific publication on Bayesian Modelling of volleyball game outcomes with professor I. Ntzoufras and Dr. S. Drikos. 

 
Bill Mexias is the COO of FSI - Fantasy Sport Interactive Limited, a Fantasy Sports software and Performance Betting services provider, helping betting operators offer a skill game to their audience.  Bill also holds an MSc in Risk Management from London’s Cass Business School and is a graduate of the Economics Department of Athens University of Economics and Business. His previous roles include risk management positions in investment banks as well as C-level positions in startup companies.

 
Ioannis Ntzoufras is a Professor of Statistics and the President of the Statistics Department of Athens University of Economics and Business (AUEB). He is a founding member of the AUEB Sports Analytics Group alongside Dimitris Karlis. He has acclaimed scientific activity in fields like Bayesian Statistical Methodology, Computational Statistics, Biostatistics, Psychometry and Sports Analytics.

 
The Athens University of Economics and Business Research Team - AUEB Sports Analytics Group was founded in 2015 by professors Ioannis Ntzoufras and Dimitris Karlis. Its members include distinguished members of the Sports analytics community, such as Stefan Kesenne (University of Antwerp & Leuven), Leonardo Egidi (Trieste University), Ioannis Kosmidis (Warwick), Konstantinos Pelehrinis (Pittsburg), Nial Friel (UCD) and Gianluca Baio (UCL), as well as the former coach of the National Volleyball Team of Greece, Sotiris Drikos. The Research Team organizes a series of annual conventions under the title of “AUEB Sports Analytics Workshop” (5 in total) while in 2019 they organized the international MathSport 2019 conference hosting 200 scientists from all over the world. The Team has a series of important scientific publications in the field of Sports Analytics. 

FSI – Fantasy Sports Interactive is an awarded Fantasy Sports Software company operating in the global iGaming and betting industry. The Innovation FSI introduced to the betting and gaming industry and for which it has become known for, is the “Fixed-Odds-Fantasy”: a cutting-edge fusion of Fantasy Sports and sports betting (Sportsbook), based on big data and the philosophy of Fantasy Sports as a sports performance index. Since 2019, when FSI presented their innovation to the industry, they have proceeded to significant agreements for the provision of the company’s original Fixed-Odds-Fantasy markets to renowned gaming companies operating both on B2B and B2C levels, such as OPAP, Scientific Games, Novibet, Hollywoodbets and more. For more information please visit: https://www.fantasysportsinteractive.co.uk/ 



CONTACT US

  • sales@fantasysportsinteractive.co.uk
  • 296 South Main Street, Suite 400 , GA 30009 , Alpharetta, USA
  • 393 Lordship Lane, N17 6AE, London, UK
  • 115 Kifisias Avenue, 115 24, Athens, Greece
Copyright © 2020 Fantasy Sports Interactive. All Rights Reserved