4-4-2 vs. 4-2-3-1 and the defense: A statistical look at shots allowed by formation

Charlie Crowhurst

Recently, there has been some discussion on the site about Bolton’s defense and whether the Trotters should be playing a 4-2-3-1 or a 4-4-2 with regards to the team’s defense. So, I decided to take a look at the shots and goals allowed over the last six league matches to see what the stats could tell me. I plan to do this in two parts, this first part will examine the shots allowed and the accuracy of those shots based on Bolton’s formation.

Before I delve into the numbers I’d like to make a few disclaimers.

First, due to living in the upper Midwest of the U.S. I was not able to actually watch any of these matches. Data on lineups, time of player substitutions, amount of stoppage time, and shots was taken from BBC Football match reports or ESPN’s Gamecast.

Second, I based my decision on what formation Bolton was in on the players on the field. If I saw a lineup like Sordell, Eagles, Chungy, and Pratley, then I classified the formation as 4-2-3-1. If I saw a lineup that was Steve de Ridder, Mark Davies, Craig Davies, and Kevin Davies, then I classified the formation as 4-4-2. The 10 men classification indicates the time in the Forest match after Marcos Alonso was sent off.

Third, the distance of a shot was determined by looking at the shot chart graphic provided ESPN’s Gamecast and comparing it to the text descriptions on BBC football’s Live text commentary. The two seemed to line up well.

Fourth, all the data and analysis comes with a small sample size warning. Six matches of data is a fair amount, but only represents about 13% of an entire season. As such short term variances may cause the data to paint a slightly different picture than what would be born out over the course of a season. I chose to only do the last six league matches, because I felt it was indicative of when the club had kind of settled in after the additional Craig Davies and the return to health of some players. So, I felt it might be indicative of how the club might perform going forward.

If you’re still with me after all of that….

Time for the tables and graphs.

First up a table of the number of shots allowed and on target by formation.

Taking a broad look at the data the first thing that jumped out at me is despite the fact that Bolton started four out of the six matches in a 4-2-3-1, the overall time spent in a 4-2-3-1 versus a 4-4-2 is about even in total. Next, Bolton has allowed 48% more shots while playing in a 4-4-2 then when playing a 4-2-3-1 (last match’s 22 shots from Derby makes up more than half this total). However, opponents have been noticeably more accurate in their shooting against Bolton’s 4-2-3-1 (this could be a function of Forest’s 9 on target out of 10 against the 4-2-3-1). The data seems to suggest that Bolton’s 4-4-2 offers opponents more opportunities to shoot, but opponents seem to make better use of their chances against the 4-2-3-1. Also, as a note of comparison, the league wide average for percentage of shots on target is 49.91%. So, Bolton’s 4-4-2 is experiencing a slightly below rate of shots on target, while in a 4-2-3-1 Bolton opponents getting shots on goal at a much greater rate than league average.

To check to see if this was a case of opponents taking more long shots against the 4-4-2, I broke down the total shots against and shots on target for each formation by distance categories: Inside the 6 yard box, Outside the 6 yard box but Inside the 18 yard box, and outside the 18 yard box. Looking at the data this way gave me the following graph.

The graph shows that when playing a 4-4-2 Bolton has been giving up more shots from all areas of the pitch, not just longer shots. From a distribution standpoint, the 4-4-2 shots are skewed a bit to the right, which may account for the slightly below league average rate of accuracy. However, looking at the distribution for the 4-2-3-1 shots, there does not appear to be a noticeable culprit for the well above league average accuracy of Bolton’s opponents. Opponents have managed to get the same number of shots on target from outside the penalty box against both formations, but did so in nine fewer attempts against the 4-2-3-1. So, the type of shot being taken by opponents does not appear to be the reason for the difference in accuracy. If anything, the graph seems to suggest that Bolton have been unlucky in their 4-2-3-1, or perhaps that they are more prone to mistakes that gift opponents better chances. However, just by looking at the numbers there is no solid evidence to suggest that there should be a discrepancy between the percentage of shots on target between the two formations based on the distance of the shots taken by opponents. The aspect of the graph that I find most worrying is the number of chances that the Trotters have allowed inside the 6 yard box while playing a 4-4-2 (it works out to be about 1.64 chances per 90 minutes) because shots inside the 6 are often of the can’t miss variety that lead to goals.

At this point, I think it is fair to say that when Bolton have played a 4-4-2 they have afforded their opponents more opportunities to take a shot, especially inside the 6 yard box. However, whether because of luck or some other combination of factors, Bolton’s opponents have been far more efficient against the 4-2-3-1. So, at this point, I do not think it is safe to say definitively which formation is better for Bolton’s defense. However, as a believer in variances exhibited in small sample size and regression to the mean, I would imagine that the accuracy number against the 4-2-3-1 would start to come down in the future, based solely on the numbers. This would tend to make me think that the 4-2-3-1 would end up being better for Bolton’s team defense, but without watching the matches to attempt for factors that are not seen in the data, I would not be comfortable saying this definitively at this point.

Next time, I’ll take a look at the goals Bolton have allowed by formation, as well as rate stats like goals per shot, and goals per shot on target.

Trending Discussions