Tuesday 3 February 2015

From Raheem Sterling to Ryan Giggs.

At the age of 38 Ryan Giggs played 1480 minutes of Premier League football for Manchester United. A year later he managed only 1169 minutes and in his final season just 487 minutes. A remarkable footballer, but also one in decline.

Meanwhile, as Giggs' career was moving towards a managerial role, 17 year old Raheem Sterling's at Liverpool was moving in the opposite trajectory. 28 minutes of playing time in 2011/12 became 1752 in 2012/13 followed by a further leap to 2227 last season. A talent full of potential, and one that is currently in the ascendancy.

Regardless of their ultimate standing in the game, the career course for these two players over the last three seasons clearly demonstrates the fate of many sportsmen. Physical maturity coupled with greater experience initially brings improvement, denoted by increased playing time, but as the former begins to decline and the latter plateaus, even the most talented drop to the bench.

The winnowing of talent presents a problem at the extremes of playing age if we try to use the performance of these remaining talents to define the likely abilities of these older age groups. Giggs accounts for a quarter of the outfield players to have played at the age of 40 since 2002. Teddy Sheringham is another and Dean Windass and Kevin Phillips complete the list, although neither played very often.

So if we look at the just the performance of 40 year old outfielders, it is proportionally top heavy with very good players. Virtually all the players who could have played in the EPL at the age of 40 since 2002 have dropped out of the game.

However, if we use Giggs' falling playing time as a proxy for his decreasing ability to contribute to a Premier League team, his inevitable age decline is clear. Even if it is falling from greater heights than most other Premier League players.

This change in playing time as a player ages does open the way to chart the typical rise (in Sterling's case) and fall (in the older Giggs' case) in a Premier League player's capabilities, partly devoid of the surviour of the best which inevitably biases the smaller sample of older players.


This alternative route to chart a player's age curve is perhaps best illustrated by the case of Premier League goalkeepers. Undoubtedly, keepers appear to peak slightly later than other positions, where physical endurance and speed may be more important. And some can play on into their late 30 or early 40's.

But as with Giggs, these older players are the exception and as with Giggs, they eventually see their playing time reduce as they are partly replaced by younger players approaching or reaching their peak.

In the plot above I have averaged the amount of increase or decrease in playing time for all Premier League keepers at different ages from 2002 to 2014. And the trend is clear. Keepers initially, on average, improve their playing time, peak and then begin to lose playing time as they decline.

The age at which improved playing time switches to a decline in playing time for Premier League keepers from 2002-2014 is a month short of 29 years old.

Despite the ability of some keepers to prolong their careers, and of the 85 keepers that potentially could have played on into their fortieth year, just 8 did, the average expectation for a Premier League goalie is that they will begin to show signs of a decline once they reach 29.

Similar delta curves, based on methods pioneered by Tom Tango can be produced for other positions and decline appears at progressively earlier ages, with forwards on average peaking around 25 years.

This study was undertaken with Simon Gleave, using data kindly provided by Infostrada Sports.


  1. Is playing time a good way to rate the quality of play? At least with strikers and traditional production (goals + assists), the answer is no based on Ben Pugsley's study a few years back showing that strikers peaked in their early 20s.

    Perhaps this trend reflects default behavior of clubs towards aging keepers.

  2. Did the study look at the change of scoring/assists as players age or just sum the totals for particular age groups? One precocious youngster can easily make an age group appear much better than they typically are through biased sampling.