How does the fearless Stoke City Class of 2009 compare with the swashbuckling Potters of 2015?
Who would win out if Abdoulaye Faye, Rory Delap and Ricardo Fuller et al went up against Steven Nzonzi, Marko Arnautovic and Jon Walters? Or Kevin Wimmer, Maxim Choupo-Moting and Jese Rodriguez?
We can debate our favourites – or the opposite – long into the night… and we all have enough time to do that right about now.
But thanks to the magic of football data and the minutiae of statistics we can pit each season’s team against another in a Stoke City mini-league, simulated 10,000 times to see who would come out on top or down at the bottom more often than not.
This is the task we set infogol’s Mark Taylor, the Uttoxeter-based Stoke fan and xG pioneer.
It was at the height of Tony Pulis’s percentage game that Mark knew there should be a better way of analysing matches than pundits saying who deserved to win based on who had more shots on target.
In short, Pulis’s side might have only had a few efforts at goal but they were at a high chance of conversion, while crowding in front of their own area to restrict opposition to long and ambitious punts.
We will let Mark explain the science behind this experiment in more detail below while we get down to the nitty gritty.
The champion of champions of the Stoke City Premier League era are… Mark Hughes’s 2014/15 team, the ones which in real life blew away Tottenham 3-0, smashed Liverpool 6-1, beat Arsenal, won away at Manchester City.
It was the first Stoke team to finish in the top 10 of the top flight for 40 years and, when pitted against simulated Stoke sides from five years either side, home and, in a further surreal twist, away, twice, finished top in 42 times out of 100.
Mame Diouf was top scorer in his first season at the club, Arnautovic was coming into his own on the left, Jon Walters and Bojan formed an unlikely but brilliant partnership and Steven Nzonzi pulled the strings in midfield next to the reliable force of Glenn Whelan.
Peter Crouch and Charlie Adam played big roles too, Marc Muniesa was emerging and Ryan Shawcross stood tall.
Next up comes the Stoke team of 2010/11 which reached the FA Cup final – finishing top in 19 per cent of the simulated seasons.
The class of 2016/17 is next, with Lee Grant player of the year after his emergency arrival from Derby County reserves and Arnautovic at the peak of his powers. It makes a mockery of what happened next.
Finishing bottom in 49 per cent of simulations was the team which got relegated – far and away ahead of any other side.
Arnautovic’s boots were never anywhere near filled while Hughes’s splurge on centre-halves to go in a wing-back system… while not buying wing-backs, proved costly. He was sacked by the January and Paul Lambert only oversaw two more wins – one in his first game in charge and one in his last when Stoke were already down.
How has all this been worked out?
Mark Taylor explained: “I’ve looked at how each Stoke team performed in relation to the other 19 teams who were in the Premier League that season. This accounts for years when the top six were particularly strong or the relegation teams were weaker.
“Then I’ve looked at expected goals created and allowed by Stoke in each season. A side that created lots of good chances, without taking them is rated higher and a side that conceded good chances without being punished is downgraded. They’ve just been on the right or wrong side of variance or ‘luck’.
“I’m left with an attacking rating and a defensive rating for each Stoke team.
“We have 10 teams, one from each season. I’ve put those into a league, used the ratings to estimate the chances a game will end in a win/loss or draw and played out a 36-game season for each ‘Stoke’ team.
“Each ‘team’ plays each other team ‘home’ and ‘away’ twice, so we have a similar number of games compared to a normal Premier League season.
“I’ve simulated a season 10,000 times. I’ve then tallied up how often per 100 a side finishes first, second, third and so on to 10th.
“I’ve also recorded the average points total and goal difference to get a feel of how close or far apart each team is from another.
“I’ve given an overall ranking from the best (Mark Hughes in 2014/15) to worst (Mark Hughes and Paul Lambert in 2017/18). I’ve only used Premier League data. The Europa League campaign impacted on the 2011/12 team.”
There is a catch in all this.
Mark added: “One variable I haven’t included is wage bill compared to the rest of the league to put the raw performances into more context.
“Tony Pulis had the 17th best wage bill in 2008/09 and it did creep up. Mark Hughes was around 12th/13th best by the end.”