Warning: This post is a bit more technical than my usual fare!
The most esoteric addition to the ’20 games in’ report - which the feedback I received suggested needed further explanation – was the Situation Map, so I thought I’d share a few examples of how they’re used. I created them out of a need to compactly visualise the data I’d collected on how often teams found themselves in certain match situations and how they tended to respond, which varied greatly across the Football League.
The 5 situations I chose (in descending order of desirability) were:
- Leading by 2 or more goals
- Leading by a single goal
- Trailing by a single goal
- Trailing by 2 or more goals
There didn’t seem any point in looking at margins of more than 2 goals, as these are quite rare. There are obviously only 3 ways in which one of these situations can be resolved:
- The team scores
- Their opponents score
- The match ends without anyone else scoring
The most space-efficient way I could come up with to combine this data was to adapt a ‘tree map’ (like the one used in David McCandless’ billion-dollar-o-gram) but using two separate axes rather than just the area:
- The horizontal shows the proportion of time spent in each of the 5 match situations
- The vertical shows the proportion of the 3 outcomes for each situation
I did consider making the horizontal axis show the number of instances a team was in each situation rather than the time spent in it, which would be a ‘purer’ comparison, but using the latter gives an additional insight into how long each situation is maintained. Let’s look at an example to see how this can be interpreted – I’ve chosen Peterborough as their map illustrates its use quite neatly:
So we have the goal difference for each of the 5 situations across the top and the 3 outcomes down the left side, both colour-coded with blue being good and red bad, with the opposite sides in each case showing 10%-sized notches to help quantify the size of each segment. The darker shaded blue and red segments show goals that affect the outcome of a match, i.e. where one team equalises or takes the lead.
Here are the main observations which can be derived from this:
- The leftmost column comes up to the first notch, denoting that Peterborough spend 10% of the time losing by 2 goals or more, which obviously isn’t ideal but at least it’s the thinnest column.
- The next column is a bit more depressing – the small blue slice at the top tells us that when Peterborough go a goal down they rarely score an equaliser (this happens in just 12% of these situations). If you look at their statistics in more detail, you’ll see that even the goals they scored here were only temporary respites: they’re the only Football League side yet to take any points from losing positions.
- The broad middle column is redder than it is blue, showing that in almost 60% of cases when they’re drawing, Peterborough will concede the next goal. What’s interesting here is how slim the grey section denoting neither team scoring is, which shows how rarely the scores stay level. If you look at the Championship table you’ll see that Peterborough have only drawn one match.
- The fourth column shows that once Peterborough go a goal up they tend to do well, going on to score again in almost 60% of cases.
- The final column is surprisingly wide for a team so low in the table – in fact, no Championship team has spent more time 2 or more goals up than Peterborough’s 15%, which is really quite extraordinary. They’ve been more than a goal up on 17 separate occasions so far this season.
For comparison, let’s take a quick look at a different team – Brentford:
- You can see straight away that Brentford spend a lot less time trailing by 2 or more goals, and the absence of a red section in this column means that they’ve yet to find themselves more than 2 goals behind in a match this season (i.e. they’ve never conceded in a situation when they were 2 goals down).
- The second column is very interesting and quite unusual: 69% of it is blue, showing that Brentford will equalise in roughly 7 out of every 10 situations where they fall a goal behind. In fact they’re one of the best sides in the Football League at coming from behind, taking an impressive 43% of available points from losing positions.
- It’s also worth pointing out how much longer they spend drawing than Peterborough: 57% of the time, which is incidentally one of the highest proportions in the Football League.
- While they’re great at scoring from losing positions, you can see that they find it a lot harder to build on a lead of their own – the fourth column shows that they score the next goal in less than 20% of the situations in which they’re a goal up. If you look at their results this season you’ll see that there have only been 3 occasions when they’ve won a match by more than 1 goal.
Hopefully the examples I’ve given illustrate the potential use of these graphics. They can help you to understand how a team is likely to respond in a given situation and pack quite a lot of information into a relatively small space. I don’t expect them to be everyone’s cup of tea, not least because taking the proportion of points gained from winning or losing positions is perhaps a much simpler way of measuring this, but it’s not nearly as much fun.
EDIT: I’ve had to revise the Brentford section slightly as I realised that there was some duplicate data throwing the numbers off.