Premier League Goalkeeping: A 10 Season Analysis
NFL and football fan. I've seen my two favourite sides, Stoke and the San Diego Chargers play at the new Wembley....and both lost.
Which clubs have had the best keepers over the last decade of the Premier League? Which have had the worst? Today on the blog Mark Taylor assesses goal keeping talent in England's top flight across the past 10 seasons.
Football at the most basic level boils down to a contest between an outfield player, most usually a striker and the opposing goalkeeper. Understandably, most of the attention in this game defining contest falls on the goal scorer, but the prevention of a goal has a near equal importance in deciding how successful a team is over the course of both a single game and an entire season.
Goal keepers can often perform at the highest level for longer periods of time than their outfield colleagues, so a team which manages to find a top class keeper can often reap rich rewards. Firstly, from on field success and potentially from lucrative transfer dealings as the value of an excellent keeper is increasingly being recognised.
Adjusting For Randomness
Save percentage for both individuals and teams over a period of time currently provides the most accessible route to quantifying which teams have excelled in their choice of keeper. Shots will obviously vary in their potency, but if we collect enough data points the hope is that keepers or teams will roughly face a similar level of challenges.
We can further refine the usefulness of save percentage by allowing for the random variation from a keeper’s true ability that we can expect to see in various sized samples. A group of fair coins each has the same innate ability to produce a head, namely 50%. But it would be surprising if 20 such coins were tossed 20 times each and every coin recorded 10 heads for a success rate of exactly 50%.
Random variation would sometimes see one coin record an “impressive” 15 out of 20, while another may record a below average score. It is tempting when looking at such results to conclude that some coins are more talented than others, when the difference is merely down to random variation. In short, this distribution of successes around the expected average gives an illusion of talent where none actually exists. In excel, the =rand() function which will generate a random number between 1 and zero, combined with the f9 key to repeat the operation provides an excellent shortcut by simulating numerous coin tosses to confirm this illusion.
If we now repeat the coin toss, but add a few coins of real, greater ability, by deeming a head as the outcome if the randomly generated number exceeds 0.3 instead of 0.5, we will now generate a different distribution of successful heads. This new distribution is a result of not just the familiar random variation, but also a real, actual difference in talent between the coins.
In essence, when we use such statistics as save percentage or indeed shot percentage, we are analysing an imperfect version of a coin toss experiment with biased coins. We need to account for random variation within the sample and any further deviation from that expectation can be tentatively attributed to “bias” or differing levels of goal keeping talent that exists between the competing teams.
The Best and Worst Shot Saving in the Premier League
Defences can and do influence the quality and quantity of shots at their goal and some teams, most notably Stoke City, are excellent at blocking efforts before they reach their intended target. However, once a shot is on target, it is invariably the job of the keeper to save it.
To extend the sample size and to focus on which teams are particularly adept at acquiring goal keeping talent, I have taken every shot on target faced by every team which played in the Premiership over the last ten completed seasons. I have compared the cumulative save percentage for all 35 sides which played at least one season of Premiership football and attempted to account for the expected random variation to produce save percentages which are more indicative of the true level of talent each side had between the posts over the last ten years.
The two tables below list the top and bottom ten teams sorted by goalkeeping talent since the 2002-2003 season through to the end of the 2011-2012 season.
Top 10 Shot Saving Premier League Clubs 2002-2012
|Team||Save Pct||Saves per 240 Shots on Target|
Typically a Premiership side will face 240 on target shots over the course of a 38 game season, so I’ve also included the number of such shots which would be saved on average given each team’s predicted save percentage.
Manchester United head the table, as they have regularly done in the wider Premiership contest, indicating that even with a relatively large turnover of keepers, they recognise the importance of a top shot stopper and more often than not manage to find such a player. Less fashionable sides, such as Stoke, Middlesbrough and Fulham are also represented in the higher reaches of the table.
Bottom 10 Shot Saving Premier League Clubs 2002-2012
|Team||Save Pct||Saves per 240 Shots on Target|
It is noteworthy that certain keepers have had dual careers at various highly placed sides, perhaps indicating the quality of their raw shot stopping abilities. Tim Howard spent the early part of the decade at table topping Manchester United before moving on to third placed Everton. Edwin Van der Sar saw action at firstly Fulham (7th) and latterly United. Thomas Sorensen, acquired from Aston Villa (10th), was Stoke’s (5th) first choice when they returned to top flight action and Mark Schwarzer played at Middlesbrough (8th) before moving onto Fulham (7th). Antti Niemi is perhaps a lower profile name who saw playing time at Southampton (6th) and then Fulham (7th) before injury curtailed his career.
Save percentage is the statistic most associated to the individual keepers, but other factors, such as the quantity of on target shots faced by a team is much more team dependent. Manchester United’s collection of goal minders, as a group have proved league leading in terms of saving on target shots, but an overall defensive effort has also allowed a below league average number of just over five shots on target per match over the decade. Not only do they have good shot stoppers, but they also restrict the number of potential scoring opportunities allowed to their opponents, leading to an excellent overall goal conceding record. This is a trait they share with a Chelsea team dominated by the presence of Petr Cech.
Perennial top four sides, such as Arsenal and Liverpool also face lower than average numbers of goal attempts, which goes some way to compensate for their less than impressive collection of shot stoppers since 2002. Based on shot stopping ability alone, Liverpool were ranked 18th and Arsenal a lowly 22nd out of 35 teams.
Teams such as Stoke, Fulham and Middlesbrough slide from the peak in terms of goals conceded by allowing above average number of goal attempts. However, this shouldn’t detract from the quality of their keepers, with Stoke City’s present number one, Asmir Begovic currently proving a big money target for England’s high spending elite.
Read more of Mark's work on his The Power Of Goals blog
And follow Mark of Twitter: @MarkTaylor0
You must be logged in to post a comment! Sign up + or log in in the top right corner.
Skill? Randomness? Stats? Maybe 10-15 years ago... Look at some of the last European Cup matches: everyone knows that Napoli don't want to go on and that Barcelona don't had to win in Milan for political reason... or you're still thinking that Messi had a bad night??? ODDS ODDS ODDS and nothing else!
Very relevant point Mark, especially considering the proliferation of more often than not, meaningless data these days.
One of the most important parts of trying to understand how good or otherwise players or teams are involves separating the randomness that exists in any process from the signal or skill. Some look at a stat and see only skill, which inevitably leads to them greatly over or under rate the talent they think they see. Understanding that universally ignored fact, I would humbly suggest is hardly useless.