On the points teams have in relation to 17th place, and the frequency with which they’re relegated

I’ve been tweeting a lot of statements like this in the last week:

Now I’m going to present the data I’m basing those on. Basically what I’ve done is taken each Premiership season and rather than using how the table looked on a given date I’ve generated the table after each team has played a given number of games. This is an interesting way of looking at things because the ’13-14 season is the first in my memory where it’s happened on a regular basis. I’m not sure whether it’s the result of a conscious effort by the FA, but in my view it’s much cleaner and easier to look at.

As a West Ham fan I’ve been interested in how the relegation fight is shaping up – it’s a race I didn’t expect us to be in and tend to think we’re better than the 17th placed team in the league, but that’s where they are and, until this weekend, they’d been in the relegation zone. As a result I began to wonder whether just how bad of a sign this was. Over half of the season has gone now and if you’re say 2 points behind 17th now you’ve got to outscore them by 4.5 points per 38 games to catch up, and 4.5 points per 38 games is a lot – it’s translates to ~2 league positions over the course of a season.

I’ve chosen to judge teams based on their point totals compared to the team in 17th place. Intrinsically I’ve tended to think goal difference was important but I looked into it and in the last thirteen seasons (my dataset) only two teams have been relegated on goal difference (Sheff United in ’07-08 and Reading in ’08-09), and only once has a team finished second in the league on goal difference (United in ’11-12). No teams have missed out on 4th place due to goal difference – meaning it has been the deciding factor <8% of the time (3 times out of 39) when it comes to deciding important positions in the league table.

The first thing I’ve done is split teams into three groups – those that have fewer points than the team in 17th place, those that have the same number of points as the team in 17th place, and those that have more points than the team in 17th place. Those are somewhat of a mouthful so from here on out I’m going to refer to them as pts<17th, pts=17th, and pts>17th respectively. I've then looked at how many teams fall into each group after a given number of matches, and how many of the teams from each group went down at the end of the season. The results are shown in the plot below.

more fewer grouped

Things I take away from this plot:

There’s a very quick rise in the pts<17th group at both the beginning and the end of the season. I think these are due to two separate mechanisms. Firstly bad teams can get themselves in a hole very quickly by not scoring points. That makes sense, although it is surprising to me just how quickly it happens. Half of the teams in the pts<17th group after 5 games are relegated, and that number rises to 60% after 7-10 games.

During mid-season there's likely to be at least one team that's down there simply due to bad luck through injuries/PDO/unfortunate refereeing decisions/bad personnel/poor management. Some of those things rectify themselves naturally, others may do so with changes in personnel in January or management at any other point in the season, and that team probably (but not always) gets out of trouble. This is, I'd suggest part of the reason for the sudden rise at the end, with the other being that teams in the pts>17th group being relegated diminish as the season progresses. That’s also not a surprise. The average of the group will move an increasing number of points above 17th as the season progresses, making it increasingly unlikely that a member of the pts<17th will catch them.

When I cut this data finer I begin to hit sample size issues but I'll leave you with one plot to demonstrate an obvious point – the more points you are behind 17th the more likely you are to go down, and the further ahead of 17th you are the more likely you are to stay up. I've only included points on this plot that consist of a minimum of five teams.

more fewer detail

So there’s noise there but the general trend and the take home messages are pretty clear – the large majority of teams that have been in the pts<17th group after 25 games have gone down, those that are 1-3 points above 17th after 28-30 games have historically gone down approximately 1 time in 8, and those that are 4+ points above 17th haven't been relegated at a rate greater than 1 time in 10 at any point during the season.


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 )

Google+ photo

You are commenting using your Google+ 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 )


Connecting to %s