How do you measure team strength?

[HoustonSoccer]

How do you measure team strength?

We all know FIFA’s methodology leaves out many other factors that influence a game. And then on paper teams may look awesome, but the game is played on the field. And as we all know many factors influence a game and there are some that are beyond a team’s control when playing in a tournament or a league. And some of these factors could be:

1) Venue
2) Weather
3) Level of Officiating
4) Injuries

And then there are others that teams can work with and influences only to the degree of preparation they make. These factors may include:

1) Player abilities and motivation
2) Team chemistry
3) Level of scouting and knowledge of the opponent
4) Preparing a game plan that maximizes the team’s strengths and
minimizes the opponent’s influence.
5) Pre-game psychology

We know from American Football many of these factors have been developed to a high art. Is there something we can learn and apply them to soccer?

The reason to question current predictions is as what we have seen in the last World Cup – it takes only one goal to overturn on all analyses. France’s defeat in its first game in WC 2002 is one such example of many.


Anyone with other ideas …

I believe if USA beats CR, they will win their group and if they lose it will be hard for them to get out of their group.

[gazzsamurai]

the most obvious way possible get weights and measure how much they can lift and count it all up then use mathamatics to work out your teams strength
duh seems pretty simply to me
and btw am joking

[HoustonSoccer]

the most obvious way possible get weights and measure how much they can lift and count it all up then use mathamatics to work out your teams strength
duh seems pretty simply to me
and btw am joking

I know you are - all I am saying is that the results of all closely placed teams are hard to predict despite the best mathematical, etc. efforts.

[Soccernethost]

Results of individual games aren't just HARD to predict -- they're impossible to predict with accuracy.

The best you can hope to do is accurate guage the likelihood of each outcome and then test your estimates over long periods of time. i.e. look at the last 100 times you said a team has a 25% chance of winning and see if they in fact won ~25 of those games.

In situations where you have enough trials and decent input variables you'll find that where models predict teams have a 20% of winning those teams almost always have a collectively lower winning % than teams the model says has a 25% chance of winning.

And past results are the best way to generate those models because they incorporate all the variables you list above. It's not perfect, but it's better than the guesswork/subjective criteria you mention above.