How The Score Affects A Team's Ambition
How does the current score impact the way teams attack and defend? Today on the blog Mark Taylor offers an analysis of Arsenal's 2010/2011 season to show us how the current match score impacts upon a team's shooting efficiency and the current limitations of simple shot on goal data.
One of the biggest and most exciting challenges presented by the copious amounts of newly available football data is to be able to apply context to the numbers. Especially as teams alter how they prioritise either their attacking intent or defensive responsibilities, depending on such things as the current match score.
Teams may choose to adopt a more defensive approach if they hold a lead or be forced into taking a more adventurous, attacking stance if they are trailing, especially as the game nears fulltime. We can begin to appreciate the extent to which teams are willing and able to adjust their “in game” strategy if we begin to look at such statistics as shooting efficiency when teams are either trailing or leading.
Game State And Attacking Ambition
If we further wish to include data collected when teams are level, we usually need to look beyond the mere score line. Drawing on the road at the Emirates, for example would be a satisfactory position for all but the very best of teams, but the home side, Arsenal would almost always find such a game state disappointing. Therefore, to include data where teams are stalemated, we need to initially concentrate on the very best sides, who are more likely to be pushing for a win even when they are still level.
Alternatively, we need to devise a method of calculating how many league points a side would expect to accrue, on average from their current game position and compare this to their pregame estimate to enable us to gauge how content they are in a stalemate.
In this post I will adopt the simplified method by looking at a leading Premiership side, Arsenal and track how the efficiency and identity of the players who were taking their shots at goal changed at different current scores. By choosing Arsenal, we can reasonably assume that they are happy to be leading, but usually eager to alter the score line if they trail or are merely drawing, even away from home.
A Case Study: Arsenal 2010/2011
Over the course of all 38 games from the 2010-11 campaign, Arsenal led for a cumulative time of just over 23 hours, were on level terms for nearly 24.5 hours and trailed for 9.5 hours. The frequency and efficiency of their shooting was noticeably different during those three distinct game phases.
Arsenal’s Shooting From Open Play. 2010-11
|Arsenal's Game State||Average Time Between Shots||Average Number Of Shots Needed To Score|
|Leading||5 minutes 40 seconds||8.2|
|Drawn||5 minutes 2 seconds||10.0|
|Losing||4 minutes 40 seconds||10.2|
When they were already leading in the match, their goal attempts became less frequent compared to when they trailed. They shot on average once every 5 minutes and 40 seconds when they led, but that time shrunk to just 4 minutes and 40 seconds when they found themselves behind.
However, the more frequent goal attempts when losing were not as efficient as their attempts when playing with a lead. They required just over eight shots to score when already ahead, but ten when they were losing and presumably facing an opponent intent on defending a lead. The respective figures when the game was tied lay between these two extremes.
So these preliminary figures suggest that Arsenal, at least in 2010-11 were frugal and efficient with their shooting when in a comfortable match situation and more extravagant, but partly wasteful when they wanted to improve the current score line.
Game State And Shot On Goal Efficiency
We can further expand our investigation by comparing the actual outcome of every shot taken by Arsenal that season in open play to the expected outcome derived from a shot location based model. Shots from open play should better reflect the dynamics of the match in different game states, because corners and set plays are likely to be defended with similar intensity regardless of the current score.
Arsenal attempted 208 shots from open play when leading and a baseline figure for a typical Premiership side would see just less than 20 goals scored. The 2010-11 Gunners, as befits a top side actually scored 24 times, so they over performed by 20%.
When trailing or drawing their 324 attempts would have likely resulted in 32 goals from an average side and Arsenal scored 34. So Arsenal’s conversion rate was still above average when trailing or drawing, but less so. A shot specific model also appears to confirm the trend of Arsenal’s declining shooting efficiency when they are trying to score while they are either behind or level.
If we finally move on to try to discover why this drop off occurs, we should first consider the obvious scenario that opponents pack their defence more when they themselves hold the lead. Unfortunately, the location of opposing players is currently beyond the scope of most mere shot location based theoretical models, so we can only guess at the impact of this likely tactical rearrangement by the opposition.
However, we can track the increase in the frequency of blocked shots in different game states and the difference would appear to be significant. When chasing the game, Arsenal’s blocked shots are 16% above the totals predicted by the base line model, but 25% below the baseline when they lead. These figures strongly suggest Arsenal face a much reduced number of defensive bodies when they attack from a position of supremacy on the scoreboard and this improves their shooting efficiency.
Other factors which should impact on scoring efficiency can be measured and the size of the open play contribution towards shots and goal attempts from presumably less skilled defenders would appear to be the most noteworthy. Defenders are fully occupied with their defensive duties when Arsenal lead and only accounted for 2% of their goal attempts from open play. But as the need for goals became more urgent, the proportion of their shot contribution rose to over 15%.
The experience of Arsenal is far from unique and these results emphasise the importance of game state context when looking at shooting efficiencies. An appreciation of how happy a side is with the current score line, especially when the abilities of each side is less extreme than in the case of games involving Arsenal, is the next small step towards better understanding the dynamics of a football match.
Read more of Mark's work on his The Power Of Goals blog
And follow Mark of Twitter: @MarkTaylor0