I’ve been guilty in these pages of making too much of a villain out of Charles Reep, the RAF wing-commander turned proto-football analytics guru. I wrote last month, in response to Bobby McMahon’s excellent piece in Forbes magazine:

McMahon goes on to answer his own question: Reep, unlike James, was undeniably a quack. And this is one of the major problems with the famous RAF Wing Commander turned amateur soccer statistician: he was indeed a pioneer, a bona fide “performance analyst,” influencer of Charles Hughes, Graham Taylor, even Sven Goran Eriksson. He was also fundamentally misguided, and arguably helped set English football back a generation, if not more, whilst sowing the seeds of media skepticism surrounding football analytics that remain to this day.

In some ways, this is unfair, particularly in light of Reep’s work on determining the average ratio of goals scored to shots taken. The average across all teams and all leagues almost always regresses to 9-10 shots for every one goal.

This ratio has held until this day. And it’s integral to one of the most important questions in analytics: what elements make a team superior over a statistically meaningful sample size? We know a little about some metrics that do not regress to the mean for successful teams, like TSR. And we also know that shots are evenly distributed over time, shots on target less so, and goals least of all. As Chris Andersen writes, “Understanding why and how this slippage occurs should be important questions for any budding analyst.”

Well, Reep in some ways was a pioneer with regard to this crucial question. Reep accepted that 1 goal to 9-10 shots ratio was the mean to which even the most effecient goal-scoring teams regress. It was static and observable over time. As Tim McGarry and Ian M. Franks observe in their essay, “The science of match analysis”, on Reep’s approach:

The use of direct play by some teams is based on the negative binomial distribution of passing sequences, as well as the remarkably stable goal/shot ratio. Thus, on the basis of probability it would seem that, in the long run, an increase in the number of shots at goal would be expected to boost the tally of goals. In addition since most shots at goal arise routinely from sequences containing few passes, the playing of direct passes into and within the scoring zone may be predicted to maximize the population of shots from which goals are scored.

There are some obvious problems with this approach. While the ‘Reep ratio’ of goals to shots has proved “remarkably stable,” the passing sequence to goals statistic can’t itself be isolated from a broader tactical approach. Meaning a team that consciously tries to score from a few passes for its own sake, without additionally successfully controlling the ball in all areas of the pitch as a matter of course, isn’t likely to beat tiki-taka Barcelona any time soon. Moreover, taking shots for shots sake after a certain number of passes in order to up the percentages isn’t going to work. Some shots, after all, taken from certain positions in sequences, are more likely to produce a goal than others.

Still, at least Reep understood that the goal to shot ratio is a deep-lying constant, a statistical anchor, and that if it could be affected in some ways, it would give a team an advantage over and above simply buying excellent players (MONEEEEEYBALLLL!). It was Reep’s inability to see his own blindspots that has led to his legacy as staunch, McNamara-like technocrat.

But, as Simon Gleave notes in a post today, his legacy led to some crucial developments in how the game is played, for better for worse. Gleave writes on John Beck’s tenure at Cambridge United:

It was the 1990/1991 season though in which Beck’s tactical ideas began to truly take shape. It was rumoured at the time that he employed a statistician although the likelihood is that this probably meant that someone simply counted various aspects of the team’s play. All good ideas have a level of simplicity to them and Beck’s were no exception: the more times you get the ball into the box, the more likely you are to score. Balls played into the box from midfield are easier to defend so use wingers, get the ball to them quickly, get them to cross early and get bodies into the box to get on the end of those crosses.

Familiar of course, but Beck was not employing Reep’s theories for their own sake. This approach led to some teams realizing the enormous advantage of set-pieces as well. We may not find this approach aesthetically pleasing, but it’s a simple method to attempt to manipulate your club’s chances without having to open up an academy and attract the world’s best players.

Comments (3)

  1. Same exact concepts in hockey shot analysis. PDO always regresses to 1000 with a big enough sample size. Also the more comparable you make the shots your using (outside the box, inside the box, quadrants in the box, etc..) the quicker the shooting percentage regresses to the mean.

  2. All this statistical nonsense is confusing me… I’d rather listen to Alan Hansen and Alan Shearer make definitive conclusions from their brilliant perspectives instead! *la sarcasme

  3. I have a question which seems to be pertinent given the MCFC/Opta data dump…how much does the analysis of the Premier League stats translate to other levels of football, I.e. non-league, academy, women’s football, etc?

Leave a Reply

Your email address will not be published. Required fields are marked *