In a previous article, I explained how to calculate expected goals and build your own simple expected goals (xG) model using data that is widely available on the internet. Whilst this xG data will provide you with a greater level of insight than merely looking at goal or shot figures, it is still limited as it treats all shots (big chances aside) from an area of the pitch as worth the same value.
How does speed of attack impact the chance of scoring?
The speed of the attack has a massive impact on how likely a shot is going to be scored. Fast breaks usually occur during counter attacks, when the opposing team’s defence is not set properly. A shot in such circumstances will have a better chance of being converted than if a side has two banks of four players lined up to minimise the routes to goal.
The good news for bettors is more and more websites which provide expected goal data are appearing, and more importantly, some break the figures down depending on a variety of factors.
One such site is understat.com, and this article will take a look at some of their data to see what insights it can provide and whether they can assist with your betting choices. For the purposes of this study I have focussed on the 14 teams who have been in the Premier League for the last three full seasons, and are still there in 2017/18. These teams have the largest understat dataset available and will be of most interest to bettors as they remain in the top flight.
The earlier comment regarding the impact that the speed with which a team attacks has on the quality of a subsequent chance is solidified by the data.
Attack speed
|
Shots
|
Goals
|
Conversion
|
Percentage of Total Shots
|
xG
|
xG/Shot
|
Fast
|
2551
|
383
|
15.0%
|
6.2%
|
392.33
|
0.15
|
Standard*
|
10938
|
1262
|
11.5%
|
26.5%
|
1255.03
|
0.11
|
Normal
|
25821
|
2410
|
9.3%
|
62.7%
|
2468.46
|
0.10
|
Slow
|
1893
|
125
|
6.6%
|
4.6%
|
108.12
|
0.06
|
Total
|
41203
|
4180
|
10.1%
|
N/A
|
4223.94
|
0.10
|
*"Standard" attack speed refers to all shots from a penalty, set piece or corner.
As we can see, shots produced from fast attacks are rare, but they are comfortably the most potent attack speed. For bettors to potentially use this information to their advantage, they need to know how different teams fare compared to the average. This is relevant both in terms of how potent their fast attacks are and how often they do them. Here are the team-by-team figures, sorted by expected goals per shot.
Premier League attack speed xG data
Team
|
Shots For
|
xG For
|
xG Per Shot
|
Team
|
Shots Against
|
xG Against
|
xG Per Shot
|
Man United
|
73
|
18.1
|
0.248
|
West Brom
|
86
|
9.9
|
0.115
|
Man City
|
110
|
23.0
|
0.209
|
Leicester
|
91
|
11.7
|
0.128
|
Arsenal
|
99
|
20.1
|
0.203
|
Tottenham
|
101
|
13.1
|
0.129
|
Leicester
|
148
|
27.2
|
0.184
|
Liverpool
|
73
|
9.6
|
0.131
|
Swansea
|
51
|
8.2
|
0.161
|
Swansea
|
85
|
11.8
|
0.138
|
Chelsea
|
117
|
18.2
|
0.155
|
Arsenal
|
94
|
14.1
|
0.150
|
Everton
|
92
|
14.3
|
0.155
|
Average
|
-
|
-
|
0.154
|
Average
|
-
|
-
|
0.154
|
Southampton
|
80
|
12.4
|
0.155
|
Liverpool
|
98
|
13.9
|
0.142
|
Everton
|
86
|
13.5
|
0.157
|
West Brom
|
48
|
6.7
|
0.139
|
Stoke
|
106
|
17.0
|
0.160
|
Tottenham
|
120
|
16.0
|
0.133
|
Chelsea
|
74
|
11.9
|
0.161
|
Stoke
|
102
|
13.2
|
0.129
|
Man United
|
67
|
10.8
|
0.161
|
West Ham
|
104
|
12.7
|
0.122
|
Man City
|
62
|
10.1
|
0.162
|
Crystal Palace
|
86
|
10.4
|
0.121
|
West Ham
|
101
|
16.8
|
0.166
|
Southampton
|
94
|
10.9
|
0.116
|
Crystal Palace
|
103
|
17.1
|
0.166
|
Bear in mind that the above figures are taken from three seasons, or 114 matches, so most teams don’t even average one fast attack for or against per game. But we can see which teams have been strongest and weakest at both ends of the pitch in terms of chance quality. The data can also be used to see how teams fare in terms of converting their fast attack shots.
Premier League attack speed conversion data
Team
|
Shots For
|
Goals
|
Conversion
|
Team
|
Shots Against
|
Goals
|
Conversion
|
Man United
|
73
|
21
|
28.8%
|
Swansea
|
85
|
7
|
8.2%
|
Everton
|
92
|
21
|
22.8%
|
Leicester
|
91
|
9
|
9.9%
|
Man City
|
110
|
23
|
20.9%
|
Man United
|
67
|
7
|
10.4%
|
Arsenal
|
99
|
20
|
20.2%
|
West Brom
|
86
|
9
|
10.5%
|
Leicester
|
148
|
28
|
18.9%
|
Arsenal
|
94
|
10
|
10.6%
|
Chelsea
|
117
|
19
|
16.2%
|
Everton
|
86
|
10
|
11.6%
|
Swansea
|
51
|
8
|
15.7%
|
Tottenham
|
101
|
12
|
11.9%
|
Average
|
-
|
-
|
15.0%
|
Liverpool
|
73
|
10
|
13.7%
|
Liverpool
|
98
|
14
|
14.3%
|
Chelsea
|
74
|
11
|
14.9%
|
Stoke
|
102
|
13
|
12.7%
|
Average
|
-
|
-
|
15.0%
|
West Brom
|
48
|
6
|
12.5%
|
Southampton
|
80
|
13
|
16.3%
|
Tottenham
|
120
|
15
|
12.5%
|
Crystal Palace
|
103
|
18
|
17.5%
|
Southampton
|
94
|
11
|
11.7%
|
West Ham
|
101
|
18
|
17.8%
|
West Ham
|
104
|
11
|
10.6%
|
Stoke
|
106
|
19
|
17.9%
|
Crystal Palace
|
86
|
7
|
8.1%
|
Man City
|
62
|
13
|
21.0%
|
Most teams are understandably in a similar position here as they are on the expected goals per shot table, but there are also some big movers.
On the attacking side, Everton are the only team to have moved by more than two places in the table. The Toffees had the seventh best xG per shot on fast attacks over the last three full seasons, but the second-best rate of converting those shots.
Without watching the attacking moves back on video it’s impossible to say how they scored so many, but their underlying expected goal figure suggests it won’t continue. Indeed, Everton only scored one of their first 12 fast attack shots in 2017/18.
There are eight teams who have moved by more than two places on the defensive tables, which is in part due to the range of conversion rates being more tightly packed than it is on the attacking side. The main movers have been Manchester United, who have allowed the fourth highest expected goals per shot on opposition fast attacks but have the third best rate of shot conversion.
To explain why would again require a review of video footage, but it seems reasonable to assume that having a goalkeeper of the quality of David de Gea will have certainly played a part.
Putting speed of attack data to good use
So how can we use this data? It’s key to remember that if we can access this data then so can the clubs. As an example, Jose Mourinho will know his side are strong on attacking counter attacks and Manchester City are weak at defending them, so it makes sense to defend deep and try to catch them on the break.
Shots produced from fast attacks are rare, but they are comfortably the most potent attack speed.
This, in turn, increases the probability of a low scoring match, and also that Manchester United will score first, as they will be tough to break down and can potentially exploit the opposition’s area of weakness. The recent derby match did not pan out this way, but that’s not to say the data should be ignored when selecting your bets.
As with any form of statistical insight in soccer, the information here is not without its issues and limitations. Whilst it makes sense to amass as much data about a team as possible to help inform your thinking, so much can change from one season to the next as to render any historic statistics far less relevant.
Teams change managers, stadiums and there is always a churn of players when the transfer window is open. Take Liverpool for instance; fast attacks have historically been below average in shot quality terms, yet their figure has more than doubled in 2017/18 since they bought Mohamed Salah. They aren’t having more fast attacks, but they are of better quality and more likely to result in goals when they do.
Perhaps the main issue bettors face when it comes to utilising data of this nature is that they have no way of knowing what tactics a manager will employ in advance. Fortunately, understat carry data on expected goals depending on what formations team use, so that will be investigated in a future article.