Storyboard

The modeling of football matches presents the following difficulties:

- The matches are not deterministic, that is, if they are repeated with the same teams and conditions, the result may be different.

- There are not enough games of the same teams under the same conditions that allow to model the forecast statistically.

- An eventual forecast will necessarily be statistical due to all the variability of the factors that affect the result.

>Model

ID:(59, 0)

Image

>Top

If the table of scores of, for example, the German Bundesliga championship of the year 2018 is observed, it can be noted that the score varies from a minimum of 19 to a maximum of 78 points:

https://www.flashscore.com/football/germany/bundesliga-2018-2019/

With such variability, it could be assumed that it should be possible to see the relative superiority in the different parties. This especially if matches are observed between those with the lowest score and those with the highest score.

For this reason, it is interesting to construct a table that shows either

- the goal difference between home and away, and

- the sum of goals between home and away

with the axis

- the abscissa the visiting teams, such as

- the ordered the teams visit teams

always ordered by goals achieved both as a visitor and away.

ID:(291, 0)

Image

>Top

When the local goal difference minus the visit is presented for each match of a competition so that the local teams are presented on the ordinate and the visitors on the abscissa, the following is obtained:

it would be expected that:

- in the upper left corner the difference (weak visits, strong locals) the difference was negative and large (goals visit)

- in the lower right corner the difference (strong visits, weak locals) of difference was positive and large (home goal)

Although a trend in this sense is seen, the relationship is not so clear:

- there are large negative numbers (win visit) for the case that both are strong

- there are large positive numbers (local win) in which case both are similarly strong

ID:(292, 0)

Image

>Top

When the sum of local goals minus visits is presented for each match of a competition so that the local teams are presented on the ordinate and the visitors on the abscissa, the following is obtained:

it would be expected that:
- in the upper left corner the difference (weak away, strong home) the sum was large (away scores)
- in the lower right corner the difference (strong away, weak home) from the sum was large (home scores)
Although a trend in this sense is seen, the relationship is not so clear:
- there are many smaller numbers in both corners

That is, if you observe that in the upper right corner there are several large numbers, which would imply that there is a tendency that when a strong team reaches the ball, it tends to have an option to convert. Thus

strong teams tend to convert regardless of whether their opponent is also strong.

ID:(293, 0)

Concept

>Top

From the analyzed Championship it can be concluded that:
- there are differences in strength between the teams
- the strength presents variability
- variability can be a fluctuation and/or show a trend

strength leads to
- weak teams tend to snatch the ball from each other failing to convert
- strong teams present many goals because once a team manages to seize the ball they have an option to convert

Therefore in modeling

The strength of a team should be associated with the probability that the team manages to keep the ball having an option to convert.

On the other hand, the forecast must consider

Estimate the strength (probability of keeping the ball) both in its fluctuation and variation throughout the championship.

ID:(294, 0)