A Look at the teams in World Cup 2018 through the Statistics that Matter

Discussion in 'World Cup 2018 - Russia' started by Iranian Monitor, Jul 13, 2018.

  1. Iranian Monitor

    Iranian Monitor Member+

    Aug 18, 2004
    Tehran Iran
    Many fans, especially novices to the game, give undue attention to statistics that tell very little about how well or poorly a side has performed in a match, while ignoring the most important statistical facts which have shown a far better predictor of how teams perform. One of the statistics that mean almost nothing, but are viewed as meaning something, are possessions stats. This World Cup showed how those stats are besides the point.

    Once the group stage was over, the top 5 teams in terms of possession were: 1- Spain (74%); 2- Germany (72%); 3- Argentina (66%); 4- Brazil (61%); 5- Saudi Arabia (59%)

    I think we can all agree that most of these teams weren't the best sides in World Cup 2018. In any case, the article below discusses possession football and how it fared in World Cup 2018 in more detail.

    There are other stats (e.g., shots, shots on goal etc) which are perhaps more telling, but ultimately insufficiently reliable as well. But there is one statistical tool that has shown itself quite reliable, even if not perfect. That tool, taking the name IMPECT from the group that developed it, and using a concept known as "packing" and finding which teams did better going behind the opposing side's defense compared to how many times they allowed the same for the opponent, was first used to explain Brazil's staggering 7:1 loss to Germany in World Cup 2014 where Brazil actually had more possession and shots and even tackles, but was clearly and deservedly whipped badly. That tool was tested again in Euro 2016, where the team with the best IMPECT scores won something like 33 of 36 matches.

    The article below will introduce you to the concept more.

    https://inews.co.uk/sport/football/world-cup/packing-football-statistic-world-cup-england-v-sweden/
    Incidentally, if you want to see how the teams rated based on their IMPECT scores after the group stage, here it the chart that would tell you the story. We can still quibble who was better or worse, and who deserved more or less, but the statistical tool that has shown itself most reliable to giving a picture of who is doing better or worse, found the teams rank as shown below based on their 'net' IMPECT score.

    [​IMG]
     
    Mifek, superdave and MPNumber9 repped this.
  2. Iranian Monitor

    Iranian Monitor Member+

    Aug 18, 2004
    Tehran Iran
    If you accept the IMPECT criteria as the best measure to see how a team was doing, then suddenly teams like Peru and Morocco (which won a lot of fans, including me, for their audacious play) don't look all that great. While the worst net IMPECT scores belonged to Saudi Arabia, Tunisia and Panama, Morocco ended up among the bottom 5. And Peru among the bottom 7, with Germany between Morocco and Peru. Even Mexico's IMPECT score was rather low, showing they actually were a bit lucky with the results they finished with.

    On the top of the list after the group stage, you had 1- Belgium 2- Russia 3- Sweden 4- Brazil 5- England 6-France 7-Croatia 8-Iran 9- Uruguay 10-Spain. Of course, the quality of opponents during the group stage wasn't equal for any measure to perfectly tell us who was doing comparatively better or worse, and teams might show better or worse form in subsequent games and stages. But perhaps the fact that no side -- neither Spain nor Croatia -- were able to defeat Russia in the knock out rounds wasn't surprising after all. Or the fact that Sweden made it as far as they did. Now, clearly, many would think that France in particular would deserve better than being regarded as #6, but in the group stage, France wasn't impressive either. Anyone who watched their group games can attest to that fact. But at the end of the day, the only team which finished among the top 10 in the IMPECT table after the group stage which failed to advance to the R16 was Iran. And most of the top 10 in the IMPECT table actually made up the quarterfinalists at the World Cup as well. Indeed, none of the quarterfinalists in Russia were from outside the top 10 of the IMPECT table compiled after the group stage.

    Anyway, just wanted to post this for posters like @celito and a few others who I have had disagreements with when it came to how well or poorly teams were doing at the World Cup.
     
  3. locoxriver

    locoxriver Moderator
    Staff Member

    May 22, 2005
    Los Angeles
    Club:
    CA River Plate
    Nat'l Team:
    Argentina
    The only statistic that matters: who scores more goals.
     
    Suyuntuy and 954gator repped this.
  4. Iranian Monitor

    Iranian Monitor Member+

    Aug 18, 2004
    Tehran Iran
    Actually, the statistics that matter is who got the best results. But the issue is which statistic tells you who came closest to achieving that feat? And which statistics give you the best idea of who is doing better or worse comparatively, giving a sense of what to expect?
     
  5. locoxriver

    locoxriver Moderator
    Staff Member

    May 22, 2005
    Los Angeles
    Club:
    CA River Plate
    Nat'l Team:
    Argentina
    No, because once again -- at the end of 90 minutes it comes down to who scored more goals.

    Like you mentioned, possession and shots taken in the course of a match are often false indicators of who was the better team. Likewise, it is irrelevant if one team "bypassed" more defenders than the other if by the time the whistle blows the other is the one who put the ball in the back of the net more times.

    What point is there in bypassing the rival team's defenders 100:1, having 90-10 possession, winning 10x as many tackles, taking dozens more shots at goal, etc. if by the end of the game the rival takes home the three points or advances to the following stage --- with or without merit?

    I get it: for the most part a team must dominate certain aspects of the game to win a match, but this is football and the best teams don't always win. A team can get completely outclassed in every single aspect and still come out as the winner if they just manage to outscore their opponent somehow (1 chance, 1 goal; own goal; defensive blunder, etc). It doesn't mean that they are the better team nor that they showed a stronger offense in the game, but in the end it is just that: the statistic that matters is who scores more goals than the other in 90 minutes.

    As for an idea of who is doing better or worse comparatively on a larger scale, that comes down to personal football philosophy really. I don't think judging on who bypasses rival defenders is the best way to do it, as that can also be affected by certain factors: a team who sits back with 11 players and conserves energy is obviously going to have higher rates of stopping opposing attackers in comparison to a team that plays a high, open, 3-man defense.

    Football is so even nowadays, that team X can win/lose to team Y on any given day. There isn't a staggering difference between teams (at least at the NT level, money-fueled clubs is a different story), so it really comes down to minor tactical decisions, form, and frankly luck. On another day, Brazil could have beaten Belgium, Germany could have taken down Mexico, Portugal could have eliminated Uruguay, etc, etc. There is no set formula/statistic to determine who is the "better" team.
     
  6. Kamtedrejt

    Kamtedrejt Member

    Internazionale Milano
    Albania
    Mar 14, 2017
    Hamburg
    Club:
    FC Internazionale Milano
    Nat'l Team:
    Albania
    #6 Kamtedrejt, Jul 13, 2018
    Last edited: Jul 13, 2018
    A lot people said that they were particiularly impressed by Morocco and Peru that both were very entertaining to watch and deserved to advance. But those IMPECT statistics give us an idea why both teams struggled to find the net despite their good play. They were very solid passing teams but just the last key pass or touch to find the necessary spaces was missing. Teams like Germany and Saudi Arabaia may enjoyed much posession but you sensed while watching that it was pretty ineffective.
    Those statistics are definately considerable. Out of the 16 best teams 13 teams made it through. I don't think we will see such a high figure when we look at statistics for posession and shoots.

    And yeah I knew this Packing stuff before. Sebastian Reinartz came once to German TV presented his method.
     
    Iranian Monitor repped this.
  7. Iranian Monitor

    Iranian Monitor Member+

    Aug 18, 2004
    Tehran Iran
    Some of your points are valid but are mostly besides the point. No one can dispute that in football, no matter how you look at it, the better team doesn't always win. But you can distinguish fluke results from more expected ones going by which statistical indicator comes closest to correctly telling you which side usually wins. For instance, if the side with more possession won 9/10 of matches, then you might consider a side with less possession winning a game to have been a fluke. But when you have what we saw in this World Cup, namely a situation where the side with the most possession actually losing more games than winning, then obviously having more possession has no correlation to saying that the loss suffered by a team was a fluke. In this context, the purpose of the IMPECT model is to give coaches and analysts a framework to understand the factors that have the actual highest correlation to the results we see in the field. The statistical measure they use was shown, in Euro 2016, to be extremely reliable with the teams having an advantage on the "net packing" points winning 33/36 matches. And a similar ratio is shown in this World Cup, suggesting that we now have a statistical measure that can more or less show us if a side can be expected to win or lose a match. Of course, no measure is perfect and there will be exceptions even here. But the exceptions (when they are real exceptions) point out which results are flukes and which aren't.
     
    Kamtedrejt repped this.
  8. Iranian Monitor

    Iranian Monitor Member+

    Aug 18, 2004
    Tehran Iran
    While the net ratings (i.e., the difference between offensive and defensive penetration) is what really counts the most, to give a flavor of how each of these teams did offensively and defensively, I have ranked them on that basis as well. The top 8 in the overall, net rankings, are in dark blue, the next 8 in green, the next 8 in orange, and the last 8 in red.

    Most to Least Offensive Penetration

    1. Belgium (55)

    2.
    Brazil (48)

    3.
    Russia (42)

    4.
    Germany (40)

    5.
    Croatia (39), Poland (39), Sweden (39)

    8.
    Denmark (37)

    9.
    Colombia (36), England (36), Japan (36), Switzerland (36)

    13.
    Portugal (35), Serbia (35)

    15.
    Mexico (34)

    16.
    Senegal (33), Spain (33)

    18.
    France (32)

    19.
    Korea Rep (31), Nigeria (31), Peru (31), Tunisia (31)

    23.
    Argentina (30), Egypt (30), Uruguay (30)

    26.
    Costa Rica (28), Iceland (28), Iran (28)

    29.
    Panama (27)

    30.
    Australia (26)

    31.
    Morocco (24)

    32.
    Saudi Arabia (22)

    Least to Most Defensive Penetration Allowed

    1.
    France (19)

    2.
    Iran (20)

    3.
    England (21), Russia (21), Sweden (21)

    6.
    Uruguay (23)

    7.
    Croatia (27), Spain (27)

    9.
    Argentina (30), Belgium (30), Korea Rep. (30)

    12.
    Egypt (31)

    13.
    Brazil (32), Senegal (32), Switzerland (32)

    14.
    Denmark (33)

    17.
    Australia (34), Iceland (34)

    19.
    Colombia (36), Portugal (36)

    21.
    Morocco (37)

    22.
    Nigeria (38), Japan (38)

    24.
    Poland (39)

    25.
    Peru (40)

    26.
    Serbia (41)

    27.
    Mexico (42)

    28.
    Costa Rica (43)

    29.
    Panama (44)

    30.
    Saudi Arabia (48)

    31.
    Germany (51)

    32.
    Tunisia (53)
     
    Kamtedrejt repped this.
  9. Steve Page

    Steve Page Member

    Oct 30, 2013
    Nat'l Team:
    England
    Those statistics are only relevant in the context of the individual matches they were played. No doubt they showed that England ripped Panama to pieces and Belgium did likewise to Tunisia. That has no relevance to how good Germany, Mexico, South Korea or Sweden were as they played much tougher group matches. If Germany had played Panama they wouldn't be near the bottom of the defensive rankings. Ranking nations based on the group stage is pointless as they all faced different challenges.
     
  10. Iranian Monitor

    Iranian Monitor Member+

    Aug 18, 2004
    Tehran Iran
    To some extent, what you mentioned is true. These statistics are indeed most relevant in analyzing each game and the teams within each group. But it isn't coincidence that 13/16 of the R16 teams are found in the top 16 of the rankings and that 8 of the top 10 in the rankings made the quarterfinals. Not that I agree necessarily with your example and the conclusions you draw. Germany had a tough time in its friendlies, including even against Saudi Arabia. Their problems weren't a reflection of how strong their group was. With Germany being in the piss poor shape it was, and mindless of its defense, it wasn't all that strong. The best side within that group was actually a mediocre Sweden, which consequently got a very good IMPECT score. On that measure, Mexico was actually not all that good. And since I have seen South Korea first hand and in many games, including against Iran, I have a hard time seeing them as being very good. Yet, they managed to beat Germany and looked better against Mexico and Sweden than they ever looked against Iran.
     
  11. Steve Page

    Steve Page Member

    Oct 30, 2013
    Nat'l Team:
    England
    The 16 teams that made it through to the 2nd round were not necessarily the 16 best teams. For that to be the case all the groups would have to be equal, which is impossible. The German group was arguably the most even. Nigeria were arguably one of the 16 best teams but they were in a tough group and thus do badly in this ranking. Trying to rank the 32 teams based on the World Cup is impossible.
     
  12. Iranian Monitor

    Iranian Monitor Member+

    Aug 18, 2004
    Tehran Iran
    I agree with your general point, namely that you can't rank teams properly when they play in different groups and different opponents.

    But in terms of how these teams looked compared to one another in their respective groups based on their net IMPECT standings, I will show that in my next post.
     
  13. Iranian Monitor

    Iranian Monitor Member+

    Aug 18, 2004
    Tehran Iran
    Using the net IMPECT scores, the best-to-worst teams in each group were as follows:

    Group A
    1- Russia (+20)
    2- Uruguay (+7)
    3- Egypt (+1)
    4- Saudi Arabia (-27)
    ---
    Group B
    1- Iran (+8)
    2- Spain (+6)
    3- Portugal (-1)
    4- Morocco (-13)
    ---
    Group C
    1- France (+13)
    2- Denmark (+4)
    3- Australia (-8)
    4- Peru (-9)
    ----
    Group D
    1- Croatia (+12)
    2- Argentina (+1)
    3- Iceland (-5)
    4- Nigeria (-7)
    __
    Group E
    1- Brazil (+16)
    2- Switzerland (+4)
    3- Serbia (-5)
    4- Costa Rica (-15)
    ---
    Group F
    1- Sweden (+18)
    2- South Korea (+1)
    3- Mexico (-8)
    4- Germany (-11)
    -----
    Group G
    1- Belgium (+25)
    2- England (+15)
    3- Panama (-17)
    4- Tunisia (-23)
    ----
    Group H
    1- Senegal (+2)
    2- Colombia (+1)
    3- Poland (0)
    4- Japan (-2)

    For me, the most balanced group was clearly Group H, where the spread between the best and worst in terms of their IMPECT score is negligible and all 4 teams basically hovered between +2 and -2.
     
    Mifek and Steve Page repped this.
  14. Kamtedrejt

    Kamtedrejt Member

    Internazionale Milano
    Albania
    Mar 14, 2017
    Hamburg
    Club:
    FC Internazionale Milano
    Nat'l Team:
    Albania
    Do you also have the figures for the knockout stage? It would be more telling for some teams. Russia´s stats are heavily influenced by facing such weak sides as Saudi Arabia and Egypt. It would be more interesting to see how they performed against Spain and Croatia.
     
    Mifek repped this.
  15. Iranian Monitor

    Iranian Monitor Member+

    Aug 18, 2004
    Tehran Iran
    I agree. I would like to have those stats, but don't. I will see if I can find them.
     
  16. bigsoccertst1

    bigsoccertst1 Member

    United States
    Sep 22, 2017
    #16 bigsoccertst1, Jul 13, 2018
    Last edited: Jul 13, 2018
    Interesting table:

    a) seems to lack an arithmetic mechanism for the "Net" column.
    b) all averages after 3 matches are round figures.
    c) if you sort with correct "Net" values, then R16 squad ARG falls to the lower 50% of that table. POL would ascend into the *qualified* zone.

    Subtraction errors for 8 squads: Russia, Senegal, Argentina, Colombia, Serbia, Iceland, Tunisia, and Saudi Arabia.

    Found a similar table posted by an Impect twitter account (2279 followers):


    Its tables for earlier states of group-phase are rife with "Net" subtraction errors, too.


    Any raw data for those tables? Just table pictures posted on that twitter account, so far.
     
  17. 954gator

    954gator Member

    Jun 16, 2016


    Its tables for earlier states of group-phase a rife with "Net" subtractions errors, too.

    Any raw data for those tables? Just table pictures posted on that twitter account, so far.[/QUOTE]
    Yeah not sure I'm buying that at all.
     
  18. Iranian Monitor

    Iranian Monitor Member+

    Aug 18, 2004
    Tehran Iran
    I am not sure what you mean? From what I saw, it seemed their arithmetic mechanism was rather straightforward: they subtracted the first column (number of times a side bypassed defenders) from the 2nd column (number of times that side's defenders were bypassed).

    I honestly don't have more direct information on it than what I posted. I did watch clips of a program in Iran that was analyzing World Cup performances and using the IMPECT table but otherwise what I posted is what I found.
     
  19. Iranian Monitor

    Iranian Monitor Member+

    Aug 18, 2004
    Tehran Iran
    Whether they have done their work right requires someone to go over the raw data and match it with what happened in each game and double check what they have done.

    But I like their concept in general, since I agree that a valuable statistic to tell us who was doing what in a match isn't whether some speculative shot was taken (or some simple shot to the goalie that is easily saved), nor certainly possession, but in general how many times a side was able to find itself exposing another side's defense by bypassing it as opposed to how many times its own defense was exposed. The best matrix, however, would probably still find a way to also give some sense of significant shots. Of course, at the end of the day, no stats is going to be able to replace a proper view of a game and the most any stat will do for me is to make me want to rewatch a match if the stats present a view that is very different that my impressions. There is no way, however, that stats can make me decide a game I saw played one way and make me see it completely another way.

    As such, when I look at this table, the most it does for me when it comes to some teams (e.g., Morocco and Peru) is to say that perhaps the fact that they didn't do better wasn't entirely coincidental and due to bad luck. Of course, that is assuming they have complied their figures correctly, but when it comes to Morocco in particular, I do have a clear impression that while they had a lot of activity and possession in their games, the times when they were able to truly bypass the defenses they faced were rare. Less than the oppposite. My recollection about Peru against Denmark is different but since Peru ended up with stats that showed them bypassing defenses often but also allowing their own defense to be bypassed even more, the problem with them was different.
     
  20. superdave

    superdave Member+

    Jul 14, 1999
    Raleigh
    Club:
    DC United
    Nat'l Team:
    United States
    Good stuff. But there are some math errors. Each group should have a net of zero, but the first and last group don't. Group D, too.
     
  21. Iranian Monitor

    Iranian Monitor Member+

    Aug 18, 2004
    Tehran Iran
    I simply didn't use the decimals when re-posting the packing statistics in a format that compared teams in the same group. And there were decimals because they were using averages. That might account for the fact that Group A for instance is +28/-27 or why the last group is +3/-2. Perhaps if the decimals that I just left off were used, the two numbers would match as the differences are minor?
     
    Mifek repped this.
  22. bigsoccertst1

    bigsoccertst1 Member

    United States
    Sep 22, 2017
    Its subtraction intent is clear, but obviously the "Net" column was not calculated automatically, more likely by hand.
    That's rather suspicious, when an organization centered on stats, like IMPECT, fails to publish simple "Net" subtractions (8 times out of 32).

    Something fishy about those averages too. Rounding off decimals can help *adjust* Net rankings. ARG being a perfect example of an *adjusted* rank.

    The IMPECT twitter account posted all its group-phase averages as rounded figures. Perhaps we are dealing with stats from a joke account?

    Too bad that raw data from IMPECT is unavailable.
     
  23. Iranian Monitor

    Iranian Monitor Member+

    Aug 18, 2004
    Tehran Iran
    The IMPEC table I posted was published in the article I referenced in my first post,, with analysis of the data and quotes from IMPECT about it.
    https://inews.co.uk/sport/football/world-cup/packing-football-statistic-world-cup-england-v-sweden/
     
  24. bigsoccertst1

    bigsoccertst1 Member

    United States
    Sep 22, 2017
    Yikes, that news outlet forgot to execute due diligence then: 25% subtraction errors on the IMPECT table it published.

    Obvious arithmetic errors aside, it is easy to calculate an approximation of raw data. Then again, we would be padding stats since averages were rounded off.

    Not sure how IMPECT failed to get "Net" values correct, with just 1 match played by all 32 squads: 25% errors in 32 calculations.
    [​IMG]
     
  25. Iranian Monitor

    Iranian Monitor Member+

    Aug 18, 2004
    Tehran Iran
    I don't know about their stats and the tables from the twitter account. But conceptually I like stats that look beyond shots and look at how many times a side really exposed a defense. Take Belgium-England today: I don't know the stats, but I know there were 4-5 very good counterattacking chances that made the game look one that Belgium could easily finish off, even though on many of them a poor last touch or something made them sequences that might not really appear in the stat books and fail to properly reflect on the flow of the game people were watching.
     

Share This Page