Players important that were important for their national teams
milivella
players. E.g. I've compared how many points per match Brazil made in
1957-71 in the matches with Pelé (1.61) and without him (1.40) to obtain
his contribution (+0.21). Clearly, the certainty of this kind of data is as
high as the statistical basis, i.e. as the minimum between "matches with"
and "matches without": in the previous example, Brazil played 92 matches
with Pelé, and 81 without him, so the degree of certainty is 81 (this is an
high value).
Here are the data for players with a good degree of certainty (46 or
higher):
Player Nation Years With W/o Diff Certainty
Garrincha Brazil 1955-66 1.84 1.28 +0.56 50
Zico Brazil 1976-86 1.68 1.22 +0.46 46
Pelé Brazil 1957-71 1.61 1.40 +0.21 81
Kohler Germany 1986-98 1.51 1.42 +0.09 52
Gilmar Brazil 1953-69 1.51 1.44 +0.07 94
Maradona Argentina 1977-94 1.24 1.20 +0.04 91
D.Santos Brazil 1952-68 1.45 1.47 -0.02 94
Shilton England 1970-90 1.34 1.35 -0.02 105
Matthaeus Germany 1980-00 1.36 1.50 -0.14 101
Clearly, these data are not so meaningful, because we should know:
- Who were the opponents?
- Which kind of match were they (friendly, World Cup, etc.)? Were they
played home or away?
- Who played instead of that player? (E.g. maybe Gilmar was not a very
great goalkeeper, but there weren't other good goalkeepers in Brazil.)
However, it's possibile to do a brief analysis:
- Brazil 1958-66 was a good side, but without Garrincha and Pelé they were
not a really great team.
- In Brazil, the most important players are attacking ones: not only
Garrincha and Pelé, but also Zico.
- Maradona maybe was not the great match-decider he is thought to be:
Argentina had about the same results with him and without him. Pelé seems
to be more important for Brazil than Maradona for Argentina.
Do you agree?
milivella
Sidenotes
Data from:
- Players: http://www.rsssf.com/intlp-countrywise.html
- National teams:
http://www.staff.city.ac.uk/r.j.gerrard/football/football.html (sadly, data
until 2001)
Data for players with 18-38 (too low?) as degree of certainty:
Cruijff +0.48
Amoros +0.31
Zoff +0.28
Fontaine +0.21
Van Basten +0.16
Rummenigge +0.16
Mueller +0.15
Bossis +0.14
Ruggeri +0.02
Platini +0.01
Beckenbauer -0.01
Charlton -0.03
Klinsmann -0.15
For this train of thought, see:
http://groups.google.it/group/rec.sport.soccer/browse_thread/thread/a023d3001ca83e89/
http://groups.google.it/group/rec.sport.soccer/browse_thread/thread/56ba19910712eebe/
The right formula should be:
T with P + 10 * (T with P - T without P)
So the simple difference is a good approximation.
It would be interesting (but practically impossible) to calculate the Elo
points of single players and not of teams...
PB
> - Who played instead of that player?
Because the question is who was important to his team, I see this as
part of the equation. Over the course of Eric Wynalda's international
career, the US earned 1.51 points per game with him versus only 1.14
without *. Obviously, his replacements weren't very good, but that's
part of the reason he was so valuable.
By the way, I find it interesting that the two highest-scoring names
in your table are players who had relatively brief international
careers. Perhaps the ideal measure would only look at players in their
best decade.
P
*3 pts per win, 106 games with/100 without
milivella
> Because the question is who was important to his team
Exactly: the concrete examples show that this can in no way be an
evaluation of the player's strength; it's only the importance of the player
for his team.
> Over the course of Eric Wynalda's international
> career, the US earned 1.51 points per game with him versus only 1.14
> without *.
Just for the sake of comparison with my data: with 2 points for win, it's
1.10 - 0.95 = 0.15.
Another interesting datum I 've found: Butcher gave to England (1980-90)
0.34 points per match (degree of certainty: 54).
> Perhaps the ideal measure would only look at players in their
> best decade.
Interesting. It would require a bit more work, because we wouldn't have
aggregate stats.
milivella