Last week, I had the good fortune to correspond with Richard Pollard. Our email exchange coincided with the publication of Richard’s most recent paper titled Invalid Interpretation of Passing Sequence Data to Assess Team Performance in Football: Repairing the Tarnished Legacy of Charles Reep (2019) (link).
I have been following Richard’s work since the publication of his paper on skill and chance in ball games co-authored with Charles Reep and Bernard Benjamin in 1971 (link).
I will write a much more detailed post about Richard’s work but in this brief post I want to affirm his part in the story of the emergence of the observation, notation and analysis of performance in association football.
Along with Neil Lanham, Richard is a custodian of Charles Reep’s experiences as a football analyst. Both have a vital role to play in demystifying Charles’ place in a history of ideas and practices.
Richard’s statistical insights and vision over the last forty years combined with Neil’s experience of recording oral traditions (link) make it possible to compile a rich account of their experiences in the early years of football analysis.
Neil has a book awaiting publication that, like Richard’s 2019 paper, should address some of the profound misconceptions about Charles’ work and locate it within Neil and Richard’s involvement in analysis (link).
Like Richard and Neil, I believe Charles’ work has been misrepresented and unfairly demonised. I hope to continue to share accounts of Richard, Neil and Charles’ work in the spirit of Sam Wineburg’s suggestion that each generation “must answer for itself anew why the study of the past is important” and “remind us why history can also bring us together” (link).
I did meet Charles at his home in Torpoint but did not make it to his shed. I am immensely grateful to Richard for sharing this picture of Charles with his archive of papers at the bottom of his garden. Somewhere in there is his roll of wallpaper that is a hand notation of the 1958 World Cup final that Charles notated in real time at the final. He transcribed his A4 paper notations onto a roll of wallpaper in an attempt to capture the flow of a game that had seven goals and included two goals scored by Pele (link).
There is no record of what happened to this archive. We can do much better with his legacy.
I have been focussing on their data. This led me to a revision of the data they shared as Table 2 in 1968 as Table 1 in Charles, Richard (Pollard) and Bernard’s 1971 paper, Skill and Chance in Ball Games.
As I re-read the papers, I was thinking about how we might visualise these data now.
I have transcribed the data from the 1968 paper and the 1971 re-working of Table 2 (as Table 1 in 1971) onto this Google Sheet.
I found a Football Pink report of the fabled Swindon Town v Bristol Rovers game, played on Saturday, 18 March 1950, thanks to a remarkable archive maintained by Swindon-Town-FC. Charles had to wait until the 82nd minute to notate his first goal sequence. A goal that was scored by Harry Kaye, the Swindon Town left half. It was a rare event, it was Harry’s only goal of the season.
I have corresponded with Ian Franks and Mike Hughes to learn about the story behind their 2005 paper on normalisation.
I hope to be introduced to Richard Pollard.
One of my activities later in the week was to re-read the 1971 paper Skill and Chance in Ball Games. I had not appreciated the nature of the changes to data analysis in that paper. There are some important changes to the 1968 data set. In the introduction to the 1971 paper, its statistical provenance is attested to by mention of the Greenwood and Yule hypothesis (1920). I have not seen this reference in any other football analysis paper.
The presence of Bernard Benjamin and Richard Pollard as co-authors with Charles gives the statistical aspects of the papers particular importance. Neil Lanham consulted an eminent statistician, Dr D A East a member of staff of the Statistical Laboratory at the University of Cambridge in regard to the data in his 1993 paper.
I think it is profoundly important to acknowledge this statistical underpinning of Charles and Neil’s papers.
As I have worked my way through the archive this week, my deciphering of the published papers has led me to think much more carefully about how we disclose data and operational definitions. My approach is to share data openly to enable community access to data and discussion about a shared resource.
I can see that I will need to do much more deciphering of the papers. I need to check with Ian and Mike about their paper too. I am hopeful that we might have access to their World Cup data from 1990 and 1994.
There is one point I will need to clarify with Ian and Mike. In their paper (2005: 510), they note:
The 1990 World Cup involved 24 teams and a total of 52 matches, whereas the 1994 format was expanded to 32 teams and a corresponding total of 64 matches.
By serendipity, I found reference to him in Ron Atkinson’s (2016) autobiography.
It was December 1954; I was fifteen, a ground-staff boy at Wolverhampton Wanderers watching Honved players walk out onto the pitch, a pitch that i and the rest of the Wolves staff had done our best to turn into a bog. The Daily Mail said it resembled the surface of a four day-old cattle show.
Wolves employed an analyst called Wing Commander Charles Reep, who argued that most goals resulted from three passes. It formed the basis of the theories that Graham Taylor was to use so successfully at Watford. Reep dominated Molineux to the extent that, if you played a square pass in your own half of the pitch, there would be trouble. If Stan Cullis saw it, he would go absolutely barmy.
Before the game, Cullis took us all into the Molineux Hotel at the top of the ground to watch a film of the Hungarians beating England 6-3 at Wembley the year before. We analysed the first goal: Hungary score almost straight from the kick-off.
… and then I found a Pathe News recording of the game.