Tactics have been transmitted to the players, set pieces have been prepared (or more likely not..) and opponents have been thoroughly studied. Then it’s game time and the manager sees his influence slip out of his hands. There seems to be only so much he can do during the 90 minutes. Walking impatiently along the sideline, shouting at the fourth official and handing notes to throw-in takers is probably not going to do it, but one of the obvious things he can do to try to impact an on-going game is to bring on substitute players. Little has been written on this matter, so here is my (first) try.
As there is a lot to discuss and no one is keen on a 2500+ word article (HA!) I will separate the various questions and this one will serve as an introductory post in a series of different aspects of players substitutions in the past 4 seasons of the Premier League. Here I will discuss the basic numbers of substitutions, in the other ones I will try to elaborate a bit more on the impact of substitutions.
For all analyses I will use data on all substitutions made over the last four Premier League seasons (that is since 2012-2013). In total 8,273 subs were made in those 1,520 games, an average of 2,72 subs per team per game. In 77% of the matches managers used all their subs. Here is the division:
When taking it from a manager’s perspective you see that some of the “top” managers, but more importantly top teams with bigger squads (and more money to pay the player’s match fees?), are inclined to use more subs. Here are the top and bottom 3, with some names of interest in between them:
But at what moments in the games do they use them? And what kind* of substitutions are made? How effective are they? And do subs have a better goal scoring rate than starters?
*Unfortunately I do not know the true intentions of the manager for any of the subs; the data I have only considers the role (goalkeeper, defender, midfielder or attacker) of each of the players.
In order to find answers to the previous questions I will only take substitutions that were made at the start of the second half or during the second half into account. I have also filtered out all goalkeeper substitutions. The reason is that I am looking for manager’s decisions and first half and goalkeeper subs are more likely to have been made due to forced circumstances (injuries and red cards respectively). When deleting first half and goal keeper substitutions from the data set 7,830 changes remain.
What kind of substitutions are made
The majority (60%) of the times a manager brings on a refreshment, this player holds the same role as his team mate coming off, in 17% of the cases a defensive sub is made and in the other 23% a more attacking minded player comes on:
The previous statistic obviously changes significantly when the game state of the match in course is taken into account. Logically more attacking substitutions are made when losing (37%) and drawing (22%) then when in a winning position (13%). For more defensive changes the opposite is the case.
We can also take the above from a manager’s perspective
In what minutes do managers make their substitutions
The average first, second and third substitution are made in minutes 62, 72 and 81 respectively. Here’s a distribution graph per 5-minute interval:
But again this differs per game state, as you can see below:
It would be a bit too much to publish this graph per manager here, so please use this interactive graph to make them yourselves and see how their substitution policy is.
This table show the average minute per substitution for all managers with at least 10 games: