How to avoid rank journalism: a simple expected points model

Looking at the current Eredivisie table we see RKC and Vitesse in the top four
with Heerenveen down in 13th position (prior to this weekend’s fourth round of
fixtures). It is, of course, early in the season and therefore too early to conclude
anything purely from the ranking. However, we can easily produce an expected
points table to compare to the current league table to see if the positions of the
clubs named above are surprising or not given the strength of opposition played.
The table below includes last night’s match which is not taken account of in this
article.


Here I am most interested in controlling purely for the strength of schedule and
therefore I have used the bookmakers odds on the 27 matches played prior to
Friday. I have used this website to access the bookmakers’ assessments of the
matches for the current season. The calculations are explained at the end of this
post for those who wish to see them but many readers will already know how to
do this as it is a relatively trivial exercise.

The first thing to note here is that the expected table is broadly speaking the
same as the observed. In other words the top and bottom halves of the tables
contain almost exactly the same teams in both of the rankings. ADO and Heracles
are the only clubs to appear in different halves of the two tables. As an aside
to this, the two halves of the table are known as the linkerrijtje and rechterrijtje
here in the Netherlands. This translates to left and right half and is presumably
due to the way the league table was historically displayed in newspapers and still
is on TV.

However, we are most interested in Vitesse, RKC and Heerenveen. Are their
positions in the current Eredivisie ranking really a surprise?

Vitesse are a great example of a team open to misinterpretation as they have a
rich owner and thus many people assume that big things can be expected.
However, this was the case last season too when they finished seventh, a
distance behind the top six, although this was already a big step up from their
15th position of 2010/2011. The question with Vitesse is whether they can
really challenge for a Champions League place this season. Currently they are
third which would suggest so but our expected table also illustrates that they
should have expected to be in the top four at this stage anyway. There is
little we can conclude at this stage from Vitesse’s good start to the season.

RKC are a different case because of their win against PSV on the opening day of
the season. Despite home advantage, RKC’s expected points from that match were
only 0.62 and PSV’s 2.17. This has had a major effect at this early stage of
course but RKC were still expected to appear in the top half anyway. So, RKC appear
to be overperforming but this is primarily due to a surprise victory in a single match.
The same applies to PSV’s slight underperformance of course.

Most interesting though is Marco van Basten’s Heerenveen. After finishing in
fifth position last season, many of the players, including top scorer Bas Dost
and assist king Luciano Narsingh, have been sold. Heerenveen’s current position
of 13th is therefore ripe for criticism and indeed, negative comments have
already started to appear. However, the simple expected points table had them
even lower at this stage, in 15th position, due to their tough start to the season.

I will return to this expected table regularly as the season wears on to see
how the clubs are performing in relation to the bookmakers’ expectations.

——————————————————————————————–

Methodology for the calculation of the expected points

1. Take the (decimal) odds of the three possible results of each match.

2. Divide 1 by each of the odds in order to get the chance of each result.

3. Sum these as the bookmakers will have built in a profit so the calculated
chances will add to more than 1 (in this case the figures we get show that the
bookmakers’ profit is around 7 to 8%.

4. Divide each of our original chances (from step 2) by the sum of the three
chances for each match in order to ensure that the chances of the three
possible results add up to 1.

5. Calculate the expected points by multiplying the win and draw chances for
each club by 3 and 1 respectively.

A practical example follows using the Ajax v AZ match.

1. Odds quoted were 1.64 for a home win, 3.85 for a draw and 4.85 for an away
win.

2. Dividing 1 into these (ie 1 divided by 1.64) gives 0.609756, 0.25974 and 0.206186 as the three respective chances.

3. However, this adds up to 1.075682 so each needs to be divided by this figure
to give the actual chance.

4. We now have 0.57, 0.24 and 0.19 (to two decimal places) as the chance of a
home win, draw and away win respectively.

5. Ajax’s expected points from the match were 1.95 (3*0.57+0.24) and AZ’s were
0.82 (3*0.19+0.24).

About these ads
This entry was posted in Uncategorized and tagged , , , , . Bookmark the permalink.

6 Responses to How to avoid rank journalism: a simple expected points model

  1. Stan says:

    These equations do not make sense in the walkthrough. Dividing 1.64 by 1 gives 1.64 not 0.609756. Can you re jig it please. Thanks.

  2. You are quite right. It should have been the other way around. Corrected now. Hope it is clear.

  3. Stan says:

    Thanks mate, i was beginning to think i had lost my marbles lol.

  4. Pingback: 11tegen11 | Alternative league table: The Relative Performance

  5. Pingback: Will Chivas USA & The Montreal Impact Continue to Be MLS Darlings? Extrapolating the MLS Season | The Shin Guardian

  6. Pingback: MLS Club Data Review: Progressions & Regressions | The Shin Guardian

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s