Futebol computacional

Eu duvido que esses modelos Poissonianos funcionem, mas fica 
como referencia: 

http://arxiv.org/abs/1004.2003 
The Socceral Force 

Norbert Bátfai 
(Submitted on 12 Apr 2010 (v1), last revised 22 Apr 2010 (this 
version, v2)) 
We have an audacious dream, we would like to develop a 
simulation and virtual reality system to support the decision 
making in European football (soccer). In this review, we 
summarize the efforts that we have made to fulfil this dream 
until recently. In addition, an introductory version of FerSML 
(Footballer and Football Simulation Markup Language) is 
presented in this paper. 
Comments: 20 pages, 13 figures, added FerSML 0.0.2 
Subjects: Artificial Intelligence (cs.AI); Software 
Engineering (cs.SE) 
MSC classes: 68T35 
ACM classes: H.5.1 
Cite as: arXiv:1004.2003v2 [cs.AI] 

http://arxiv.org/abs/1002.0797 
Soccer: is scoring goals a predictable Poissonian process? 

Andreas Heuer, Christian Mueller, Oliver Rubner 
(Submitted on 3 Feb 2010 (v1), last revised 3 Mar 2010 (this 
version, v2)) 
The non-scientific event of a soccer match is analysed on a 
strictly scientific level. The analysis is based on the 
recently introduced concept of a team fitness (Eur. Phys. J. B 
67, 445, 2009) and requires the use of finite-size scaling. A 
uniquely defined function is derived which quantitatively 
predicts the expected average outcome of a soccer match in 
terms of the fitness of both teams. It is checked whether 
temporary fitness fluctuations of a team hamper the 
predictability of a soccer match. 
To a very good approximation scoring goals during a match can 
be characterized as independent Poissonian processes with pre- 
determined expectation values. Minor correlations give rise to 
an increase of the number of draws. The non-Poissonian overall 
goal distribution is just a consequence of the fitness 
distribution among different teams. The limits of 
predictability of soccer matches are quantified. Our model-free 
classification of the underlying ingredients determining the 
outcome of soccer matches can be generalized to different types 
of sports events. 
Subjects: Data Analysis, Statistics and Probability 
(physics.data-an); Physics and Society (physics.soc-ph) 
Journal reference: Europhys. Lett. 89 (2010) 38007 
DOI: 10.1209/0295-5075/89/38007 
Cite as: arXiv:1002.0797v2 [physics.data-an] 


http://arxiv.org/abs/0909.4555 
Soccer matches as experiments: how often does the 'best' team 
win? 

G. K. Skinner, G. H. Freeman 
(Submitted on 24 Sep 2009) 
Models in which the number of goals scored by a team in a 
soccer match follow a Poisson distribution, or a closely 
related one, have been widely discussed. We here consider a 
soccer match as an experiment to assess which of two teams is 
superior and examine the probability that the outcome of the 
experiment (match) truly represents the relative abilities of 
the two teams. Given a final score, it is possible by using a 
Bayesian approach to quantify the probability that it was or 
was not the case that 'the best team won'. For typical scores, 
the probability of a misleading result is significant. 
Modifying the rules of the game to increase the typical number 
of goals scored would improve the situation, but a level of 
confidence that would normally be regarded as satisfactory 
could not be obtained unless the character of the game was 
radically changed. 
Comments: Contact the corresponding author in case of 
difficulty in accessing the published paper 
Subjects: Physics and Society (physics.soc-ph); Data 
Analysis, Statistics and Probability (physics.data-an); 
Applications (stat.AP) 
Journal reference: Journal of Applied Statistics Vol. 36, 
No. 10, October 2009, 1087-1095 
DOI: 10.1080/02664760802715922 
Cite as: arXiv:0909.4555v1 [physics.soc-ph] 



http://arxiv.org/abs/0803.0614 
Fitness, chance, and myths: an objective view on soccer results 

Andreas Heuer, Oliver Rubner 
(Submitted on 5 Mar 2008 (v1), last revised 20 Mar 2009 (this 
version, v4)) 
We analyze the time series of soccer matches in a model-free 
way using data for the German soccer league (Bundesliga). We 
argue that the goal difference is a better measure for the 
overall fitness of a team than the number of points. It is 
shown that the time evolution of the table during a season can 
be interpreted as a random walk with an underlying constant 
drift. Variations of the overall fitness mainly occur during 
the summer break but not during a season. The fitness 
correlation shows a long-time decay on the scale of a quarter 
century. Some typical soccer myths are analyzed in detail. It 
is shown that losing but no winning streaks exist. For this 
analysis ideas from multidimensional NMR experiments have been 
borrowed. Furthermore, beyond the general home advantage there 
is no statistically relevant indication of a team-specific home 
fitness. Based on these insights a framework for a statistical 
characterization of the results of a soccer league is 
introduced and some general consequences for the prediction of 
soccer results are formulated. 
Comments: tex-file, 31 pages,16 figures 
Subjects: Data Analysis, Statistics and Probability 
(physics.data-an); Physics and Society (physics.soc-ph) 
DOI: 10.1140/epjb/e2009-00024-8 
Cite as: arXiv:0803.0614v4 [physics.data-an]