Saturday, October 28, 2023

Another way of ranking the international top scorers

Normally you don't see anything in this blog about individual players, because they are not my main interest in NT-football. It's about the national teams for me. But recently there was a request from a long-time reader about the ranking of international top scorers, in combination with the strength of the opponents against which they scored their goals.

Especially since Portugal's Cristiano Ronaldo, the ultimate professional and prolific football player, but also the world's biggest phoney when it comes to being a team player, is breaking records, this topic is apparently hot. He's the leading all-time international top scorer with 127 goals from 203 caps. 


By the way, the numbers in this post are collected from the following sources and updated manually when data was lagging: 

- the international all-time top scorers list from wiki;

- for each player in that list an overview with all international matches in which they played (and sometimes scored) by date, maintained until July 2023 from RSSSF;

- historic elo-ratings per date.


The idea was to get a better picture of the worth of all these goals by weighing the scored goals with the strength of the opponent at the time the match was played and the goals were scored. Because in the top scorers list players from all over the twentieth century appear, the FIFA rating (existing only since August 1993) wasn't useful. Not to mention the three major calculation adaptations in 2002, 2006 and 2018 that caused severe trend breaks in the FIFA ratings.


The obvious alternative is the elo rating which has a number of advantages:

- this rating exists as soon as a national team plays its first international match. The initial rating before the first match of a new team is a (knowledgeable) estimate but within some 30 matches the rating of the team has converged to its 'real' value. The elo calculation method for the match points to exchange between both opponents in each match produces that fast convergence.

- the calculation method has never been changed, so there are no trend breaks in the ratings. All international matches since the very first one (Scotland-England 0-0 in Glasgow on November 30th 1872) are calculated with the same formula.

- the rating of both teams in a match is immediately adapted after the match.

- the elo rating simply gives a good and fairly stable insight in strengths of all teams, compared to each other.


I've restricted the overviews to all 38 players with currently 60 NT-goals or more. For each NT-match where a top scorer scored one or more goals I've determined the elo rating of the opponent right before the match was played. Each goal scored is weighed by the elo rating of the opponent so eventually an average elo rating per scored goal can be determined. The higher this average the stronger the opponents were.


Okay, enough chit chat, let's go to the numbers. 

First the habitual ranking, sorted by number of scored goals:


name

nat

goals

caps

avg gl/cap

career_span

Cristiano Ronaldo

POR

127

203

0,63

2003-

Ali Daei

IRN

108

148

0,73

1993-2006

Lionel Messi

ARG

106

178

0,60

2005-

Sunil Chhetri

IND

93

143

0,65

2005-

Mokhtar Dahari

MAS

89

142

0,63

1972-1985

Ferenc Puskas

HUN

84

89

0,94

1945-1962

Ali Mabkhout

UAE

81

111

0,73

2009-

Robert Lewandowski

POL

81

144

0,56

2008-

Godfrey Chitalu

ZAM

79

111

0,71

1968-1980

Romelu Lukaku

BEL

79

112

0,71

2010-

Neymar

BRA

79

128

0,62

2010-

Hussein Saeed

IRQ

78

137

0,57

1977-1990

Pelé

BRA

77

92

0,84

1957-1971

Vivian Woodward

ENG

75

53

1,42

1903-1914

Sandor Kocsis

HUN

75

68

1,10

1948-1956

Kunishige Kamamoto

JPN

75

76

0,99

1964-1977

Bashar Abdullah

KUW

75

134

0,56

1996-2007

Majed Abdullah

KSA

72

117

0,62

1977-1994

Kinna Phiri

MWI

71

117

0,61

1973-1981

Kiatisuk Senamuang

THA

71

134

0,53

1993-2007

Miroslav Klose

GER

71

137

0,52

2001-2014

Piyapong Pue-on

THA

70

100

0,70

1981-1997

Abdul Kadir

IDN

70

111

0,63

1967-1979

Stern John

TRI

70

115

0,61

1995-2012

Gerd Müller

FRG

68

62

1,10

1966-1974

Carlos Ruiz

GUA

68

133

0,51

1998-2016

Luis Suarez

URU

68

137

0,50

2007-

Robbie Keane

IRL

68

146

0,47

1998-2016

Hossam Hassan

EGY

68

176

0,39

1985-2006

Didier Drogba

CIV

65

105

0,62

2002-2014

Edin Dzeko

BIH

65

133

0,49

2007-

Teerasil Dangda

THA

64

126

0,51

2007-

Jassem Al-Houwaidi

KUW

63

83

0,76

1992-2003

Ronaldo

BRA

62

98

0,63

1994-2011

Ahmed Radhi

IRQ

62

121

0,51

1982-1997

Zlatan Ibrahimovic

SWE

62

122

0,51

2001-2023

Abdul Ghani Minhat

MAS

61

71

0,86

1956-1966

Harry Kane

ENG

61

87

0,70

2015-

Players in bold are still active.


Then ranked by the average opponent's elo rating per goal scored:


name

nat

goals

caps

avg elo_opp/goal

Pelé

BRA

77

92

1784,6

Neymar

BRA

79

128

1777,1

Ronaldo

BRA

62

98

1752,5

Gerd Müller

FRG

68

62

1752,4

Luis Suarez

URU

68

137

1750,4

Sandor Kocsis

HUN

75

68

1738,6

Vivian Woodward

ENG

75

53

1734,9

Lionel Messi

ARG

106

178

1714,5

Ferenc Puskas

HUN

84

89

1694,9

Miroslav Klose

GER

71

137

1660,9

Romelu Lukaku

BEL

79

112

1647,9

Harry Kane

ENG

61

87

1623,3

Cristiano Ronaldo

POR

127

203

1582,9

Zlatan Ibrahimovic

SWE

62

122

1581,3

Didier Drogba

CIV

65

105

1555,5

Robbie Keane

IRL

68

146

1552,4

Edin Dzeko

BIH

65

133

1544,4

Robert Lewandowski

POL

81

144

1506,1

Hossam Hassan

EGY

68

176

1501,6

Carlos Ruiz

GUA

68

133

1468,5

Stern John

TRI

70

115

1450,4

Ahmed Radhi

IRQ

62

121

1388,2

Godfrey Chitalu

ZAM

79

111

1376,7

Jassem Al-Houwaidi

KUW

63

83

1373,6

Abdul Kadir

IDN

70

111

1341,3

Kinna Phiri

MWI

71

117

1337,6

Ali Daei

IRN

108

148

1332,0

Majed Abdullah

KSA

72

117

1325,8

Kunishige Kamamoto

JPN

75

76

1319,8

Hussein Saeed

IRQ

78

137

1319,2

Bashar Abdullah

KUW

75

134

1310,2

Ali Mabkhout

UAE

81

111

1307,7

Mokhtar Dahari

MAS

89

142

1303,9

Abdul Ghani Minhat

MAS

61

71

1265,9

Piyapong Pue-on

THA

70

100

1178,3

Teerasil Dangda

THA

64

126

1152,1

Kiatisuk Senamuang

THA

71

134

1151,9

Sunil Chhetri

IND

93

143

1094,1


The famous Brazilian strikers Pelé and Ronaldo, together with current star Neymar, are leading this list of most 'valuable' top scorers. Golden boy Cristiano Ronaldo is way below them and -nice to see that- also quite some way below Lionel Messi. Prolific scorers Harry Kane for England and Romelu Lukaku for Belgium are sitting pretty in this list.

Not surprisingly the current (India's Sunil Chhetri, Thai Teerasil Dangda and Ali Mabkhout for the United Arab Emirates) and past Asian top scorers like Malaysian Mokhtar Dahari and Iran's Ali Daei are found in the lower echelons of this list, their opponents were and are certainly not the strongest in the world at the time of play.

You can look up your own favorite all-time NT top-scorer, of course.



About me:

Software engineer, happily unmarried and non-religious. You won't find me on Twitter or other so called social media. Dutchman, joined the blog in March 2018.

14 comments:

  1. The average elo rating is interesting to see, but I guess in some way it also just shows that some teams play a lot more bad teams than other. So in this average Pele would look worse if he had scored an additional 20 goals against some poor teams, while in reality it should probably add a tiny bit to his accomplishments. So I think it could be even more fruitful to make recalculcated, weighted totals. So that a goal against a team with an average elo-ratings equals 1 goal, a goal against a strong nation counts 1,?? based on the nation's elorating compared to the average, and a goal against a weak nation counts 0,?? . In that way you aren't punished for scoring many goals against small teams. They just don't count as much.

    ReplyDelete
    Replies
    1. That is a good suggestion. Ed, do you have an idea on how to best approach this recalculation, i.e. how to assign proper wieghtings for each goal and then calculate to weighted total?

      Delete
  2. Yeah, that was always the danger. Every simple approach will trigger a more complicated one :D
    But I certainly see your point and agree.

    A fair approach could be to weigh each goal against the over- or understrength of the opponent as Anon suggests. I could start with calculating for each ranking the average rating of the top 100 teams and then determine three classes of teams in the ranking:
    - the stronger than average teams. Let's say they have a rating of more than a standard deviation above the calculated average;
    - the teams of average strength: with a rating between minus one time or plus one time the standard deviation around the average;
    - the weaker than average teams: with a rating below the average minus one time the standard deviation.

    Each scored goal is weighted with the following weight factors:
    - goals against stronger than average teams get a weight of 2;
    - goals against average teams get a weight of 1;
    - goals against weaker than average teams get a weight of 0,5.

    I will have to play around with the standard deviations and weight factors first a bit to research the sensitivity of the approach, but I promise I will not calculate towards a desired result :)

    How does that sound ?

    ReplyDelete
  3. And come to think of it, you can also take into account the type of match. Is it a World Cup match or a friendly ? Or is it the first goal in a match or the seventh ?
    I have no information for older matches about that last aspect, but I could factor in the importance of the match, assuming a factor like the K-factor in the elo calculation.

    ReplyDelete
  4. Bonjour va t il y avoir des simulations qualification mondial 2026 zone afrique ? Merci

    ReplyDelete
    Replies
    1. Bien sûr, jetez un œil ici pour les dernières simulations de septembre.

      Delete
  5. Would it make sense to take into account the ELO rating of the NT of each scorer as well as merely the opponent? This could be a weighting based on comparison of each team's rating.

    The reason I suggest this is, for example, Neymar playing for Brazil might reasonably expect to be more likely to score against, for random example, Switzerland, than Edin Dzeko playing for Bosnia HG against Switzerland, simply because Brazil would have a greater goals expectancy than Bosnia HG.

    Yes, I think its fair to compensate that high scorers for weaker NTs generally score against weaker NTs, on the other hand Bashar Abdullah playing for KUW against KSA is likely to have less chances to score than Ronaldo playing for BRA against KSA.

    ReplyDelete
    Replies
    1. Interesting point. You suggest to not look only at the strength of the opponent, but at the same time in relation to the strength of the team with the top scorer.
      I don't have information for all older matches about xG, so I translate your suggestion to a weighting, based on the elo-difference between the team of the top scorer and the other team in each match: the smaller the elo-difference, the higher the weight. The elo-difference will range from very negative to very positive, so I will establish a small amount of difference-classes and couple a weight to each class.

      Something like that ?

      Delete
  6. Hello. Do you know UEFA futsal national team rankings starting points?

    ReplyDelete
    Replies
    1. Sorry George, I don't know anything about a futsal ranking.

      Delete
  7. Long time lurker, first time poster here! Two things I think are worth considering in addition: you mentioned the Asian players are lower on the list, and that makes perfect sense if most goals are against lower ranked teams. However, that strongly weighs their average down if, for example, they also scored hat tricks on the rare occasions they played top tier sides. I’m not saying that necessarily happened, but if 90% of their games are against low ranks and they consistently score, well you would expect the same from Pelé if he played against those teams too, right? Maybe it’s worth balancing this out with the teams they couldn’t score against? For example, maybe Pelé failed to score in a lot of top tier clashes while Chhetri had a 109% record against top tier sides. The rank also doesn’t capture the goals per game, which is a whole other story of effectiveness (some players in the world would top this list if they scored 1 goal against Argentina in their 3-cap national career, for example). I love what you’ve started here and I’d be interested to see if it could be further fleshed out 🙂

    ReplyDelete
  8. Hi long-time lurker, I plan to wait a few more days (probably until next week) to collect all interesting ideas and suggestions and then implement them and publish a second post about the ranking of these international top-scorers. Didn't expected it would stir the antheap this much :)

    ReplyDelete
  9. Hello Ed, Hope you're doing well.
    I was very excited to read the post once I saw that you have worked out on the idea I provided in the last blog. Also, glad to see such high response in this regard. So many new criterias have been highlighted to make the list more accurate. Another factor that would be important to realise is that that the number of minutes played by a player. Hope you can dig out more ways to find better approach to this topic. Cheers !!

    ReplyDelete