Home page‎ > ‎

HATTRICK - Player Speciality HEADER

In the manual, there's a part about Special Events (corner kick section) you can find this sentence:

“The higher the number of outfield head specialists in your team (your set pieces taker does not count), and the lower the number of outfield head specialists among your opponent's, the better your chances to score.”

This sentence can suggest that in every case, not only in the corner kicks events, the probability of a heading goal is in some way connectes to a comparison between the numbers of headers of a team with the number of headers of the opponent one.

Let's set

Ph - The probability of scoring of team H (playing Home)
Hh - The number of headers of team H (playing Home)
Ha - The number of headers of team A (playing Away)

it would be Ph = f (Hh , Ha)

The internet site Hattristics collects data of 317.812 matches between november 2003 and december 2008. A page of the site highlightshttp://www.hattristics.org/pub/statHeads.php
In that table you can see goals and missed chances for the home team and the away team, divided according to the number of home team headers Hh (the rows) and the away team headers Ha (the columns). In this analysis I cut off the values calculated with less than 25 matches (too little a number).

I highlighted with a
  • blue background the home team goals,
  • light blue for the home team missed chances,
  • red for away team goals and
  • pink for away team missed chances.
Under the two values there's the sum of goals and missed chances.

For example, with 5 headers for the home team and 2 for the away team we could expect 0.314 goals for the home team and 0.090 for the away one.

Let's delete from the table the missed chances data and let's focus on the goals.
The table becomes:

And now let's focus just on the home team goals:

For example let's consider the row Hh = 2 with 2 headers for the home team. We get 0.142 goals if the away team has 0 headers, 0.135 if the away team has 1, 0.135 if they have 2, 0.134 if they have 3 and so on (the latter values are less reliable, because they're calculated on a smaller sample, with a larger variance).
The "MEDIA" column shows the averages, while the "Gol per Header" shows the quotient calculated dividing the average by the number of headers.

The data suggest that:

  •  The probability of a heading goal is linearly dependent (column "Gol per Header") on the number of home team headers Hh .
  •  The probability of home team goals, given the mumber of home team headers Hh, is just slightly dependent from the number Ha of away team headers

So, more headers, more goals, in a constant way and jsut slightly dependent on the opponent's headers number.

We get an average of 0,0655 home team goals for every Header:

so, for example, if I have 3 headers in the 7 home matches of a season I expect an average of 1,38 goals.

I have to admit that in a first moment I said that the probability Ph was totally independent on the number of opponent headers. Looking back at the data I realized I missed the slight dependence.
If I set at 100% the average value I see that the values for Ha=0 are almost all above 100% and there's a general tendency to decrease while the number of opponent Headers Ha increase:

I can suppose that every opponent header Ha lowers the expected number of goals of 2%. So:

To give an idea if I have 4 headers (Hh=4), in 100 matches against team
  • with 0 headers (Ha=0), I'll expect to score 26.20 goals
  • with 5 headers (Ha=5), I'll expect to score 23.58 goals

Let's look now at the Away team numbers:
Here's the goals table

I can see that even here the values (in columns) are slightly reducing: for example I see in column 2 the values 0.217 0.120 0.117 0.105 0.101. So all in all the goals are a bit dependent on the number of the home team headers.

And again we find a linear dependence because the number of goal "per Header" is constant around his average that is 0.0551


if I have 3 headers in the 7 away matches of the season I expect 1,15 goals.

I see again that the values (in columns)

tend to decrease while the number of headers of the opponent (here the home team) raises.
If we cut off the deviant values I see that every home team header more (Hh) lowers the goal probability by around 2%.


So all in all a header will get me about 0.84 goals every season.

As a last note I wish to say that the home coefficient is around 19% more than the away one, a number fitting very well with what we found in other studies about the bonus of home playing.

Andreac (team ID 1730726 in Hattrick)
This opera by Andrea Candio is licensed under a Creative Commons Attribuzione-Non commerciale 3.0 Unported License.