The Stat Man: Predicting how the automatic promotion race in League One might play out

It won’t have escaped your notice that Sunderland have hit a bit of form lately. The timing seems perfect as the days lengthen, and we head towards the business end of the season with May coming over the horizon.

It also won’t have escaped your notice that we were, frankly, awful until Boxing Day. At that time they reached their all-time lowest league position and the clamour for a change of manager was at fever pitch.

A commonly used metric used to judge a team’s chances of automatic promotion is ‘points per game’. My old Grandad used to say ‘win your home games and draw your away games and you’ll win the league’ (I’m sure many of you heard the same from your own elders). Since 1981 and the introduction of three points for a win, this rule of thumb could also be represented by ‘two points per game’ being the target for teams seriously chasing championships or automatic promotion.

At full time in the home game against Bolton on Boxing Day, when the calls for Phil Parkinson’s head reached a crescendo, averaged over ten games, Sunderland had reached the depths of 0.90 points per game. If that were extrapolated over a full season, the resulting points haul would have been 39.6. Very much relegation form and certainly not ‘winning home games and drawing away games’ nor ‘two points per game’.

That low point, however, brought with it a watershed. A watershed beyond which the season and expectations very much changed. The turnaround wasn’t even gradual, it was sudden as is demonstrated in the chart below that shows the ten-game moving average points per game over the season right up to, and including, the home victory against Rochdale on February 11th.

As you can see, the period from game 10 to game 20 (Bolton at home) was a time of misery. In stark contrast, from game 20 to 30 has mostly brought smiles. Just as we had started to see the return of the dreaded match day apathy that became so all encompassing during the Championship season, things became as good as they were bad. So good, in fact, that - for the first time this season - our form has us gaining more than two points per game (signified by the plot being above the green dotted line). That right there is promotion form.

We are quite rightly thrilled that Sunderland are finally delivering. And what fantastic timing too. But we are not the only team to have started to deliver recently. Here is the same plot as above with the addition of the rest of the current League One top seven.

Here we can see that green (Peterborough), dark blue (Ipswich) and orange (Wycombe) all spent time earlier in the season in promotion form. Of those three teams, none remain in such form. In fact, Wycombe are technically in relegation form with a current ten-game average of one point per game. Ipswich aren’t doing much better but Peterborough, on the other hand, have recovered and are quickly heading back towards promotion form (as anyone who has seen their recent results will know).

There is a lot of information in the above chart so, to simplify the discussion, I want to isolate it to a smaller group of times and the shorter time frame shown below.

Each of the remaining five teams are showing form which has been improving at varying rates since the turn of the year (roughly game 20). This serves as confirmation that, while Sunderland have hit great form at the perfect time, so have four other teams. Namely, Coventry, Peterborough, Portsmouth and Rotherham.

Four of those form teams are already above the two points per game automatic promotion requirement, and the fifth (Peterborough) will be there soon enough.

If you’ve read my previous two articles (here and here), you will know that I have introduced a method of measuring and illustrating performance relative both to other teams and to theoretical ‘perfection’. In those articles, I used the method to compare performance in various periods in Sunderland’s 2019/20 season to that of teams automatically promoted from League One in recent years.

Next I will use that same method to attempt to assess the overall performance of the five aforementioned promotion-chasing teams and attempt to predict what could happen next.

---

Which teams might be heading for automatic promotion?

To begin to answer this question, I first present overall performance for every League One side. The chart includes data from the entire season.

Some high level (and unsurprising) observations are that Bolton, Southend, Tranmere and Rochdale have been poor over the whole season. Conversely, Rotherham, Peterborough and Portsmouth have done well.

The peloton includes Sunderland, but it’s crowded. To make things a little clearer, I present a closer look teams in the promotion mix here.

This closer look confirms that the teams producing the better performances include Portsmouth, Peterborough and Rotherham. Joining them are Coventry, Fleetwood and Oxford. A little off the pace are Doncaster, Sunderland and Ipswich with Gillingham (a potential dark horse) a little further back.

As interesting as this is, our main concern today is how the rest of the season might play out. To begin to form that picture, I next present a performance chart containing only each side’s last ten games.

If you’re a Bristol Rovers or Tranmere fan who happens to have stumbled upon this article, I’m afraid your team have been so poor recently that you’re off the scale. And spare a thought for Wycombe who - in performance terms - have been the third worst team in the league in the last ten games (only Tranmere and Bristol Rovers are worse). The wheels have well and truly fallen off at Adams Park. Our house, in the middle of our league.

Unsurprisingly, the sides I identified as doing well in the points per game charts are well positioned here too. Most pleasingly, Sunderland seem to be among the leaders now. Oxford and Ipswich, on the other hand, have fallen off the pace.

Again it’s an overcrowded picture, so let’s have a closer look at the leaders.

The dark horse, Gillingham, are still hanging on to the leaders and look a decent bet for grabbing a play off place at the death. As do Fleetwood.

Peterborough, after a period of awful form which saw their owner promise to leave the club if they didn’t make the play-offs, have improved a lot recently and certainly look dangerous as the season builds to a crescendo.

Most importantly for us, however, Sunderland - and our old foe Portsmouth - are leading the way while Doncaster, Rotherham and Coventry are tightly packed behind.

It’s not easy to see from the chart which of Sunderland or Portsmouth have the better performance stats so, using a bit of Pythagoras, the following league table is produced.

We win! Unfortunately, however, there are no prizes for producing the best performances with 14 games of the season remaining. But if the current level of performance continues, we will certainly be in the mix.

Next I present the performance chart including some arrows that show where a team was on the plot ten games ago.

The most improved (coming from off the scale) are Coventry, but Sunderland are not far behind with Rotherham (dotted arrow) just behind us.

Doncaster (black arrow) have remained steady, showing only marginal improvement while Portsmouth have taken an unusual path. Their goals ratio improved but their shots on target ratio worsened, resulting in their position, relative to perfection, being more or less the same at the beginning and end of the period.

Pleasingly, Sunderland’s arrow points convincingly in the direction of ‘perfection’. It’s unlikely such a rate of improvement will continue, but there is every reason to hope we will at least continue to head in the right direction.

Each side in the mix has about 14 games left to play. Inevitably, there will be varying difficulties of run in. We all like to look down the fixture list and make predictions about which sides have tough games and what the results might be.

That approach might be fun but it is also highly subjective. I much prefer to remove the subjectivity with some nerdy data fun.

To that end, I examined each of Rotherham, Portsmouth, Coventry, Peterborough and Sunderland’s remaining opponents and found the centroids for each. Those centroids provide a combined performance datapoint for each set of opponents. I have plotted them below.

The further an opponents’ datapoint is from the bottom left corner, the less difficult the set of opponents. On that basis, Peterborough seem to have the easiest end to the season of our teams of interest. Portsmouth seem to have the most difficult set of opponents (not a single tear will be shed in Sunderland).

To only examine the position of opponents on the plot is not sufficient however, because this does not take into account the performance of the promotion chasing team itself. For example, Coventry’s performance datapoint and that of their opponents are closer together than the pair for, say, Peterborough.

In order to produce a final ranking for ‘ease of run in’, relative performance needs to be taken into account. This is another job for Pythagoras, who gives us the result below.

It would seem that Sunderland have the easiest run in and Coventry the hardest. Excellent result! But... Rotherham, Coventry, Peterborough and Portsmouth all currently have superior points per game returns to Sunderland for the season as a whole - in other words, points on the board. So, although we might apparently have the easier run in, we’re starting from a position behind the others. I predict this is going to go very much down to the wire.

Don’t book any holidays for the weekend of May 24th just yet.