Dr Ian McHale, a statistician and Director of the Centre for Sports Business at Salford University (also Chair of the Royal Statistical Society’s Sports Section) is, of course, right to say that sport is a great playground for statisticians. After all, he co-developed the model underpinning the Official Player Rating System of the Barclays Premier League which aims to tell us which football players are best and why. He also spends time thinking through who the winners and losers are in other sports.
Just as we are all turning into Wimbledon tennis heads (I can feel the crick in my neck already), how great to be able to pit our general sense of where our favourite players are in the all time world rankings against some serious statistical analysis…right on cue, we have new research entitled ‘A dynamic paired comparisons model: who is the greatest tennis player?’ completed by Ian and co-author Professor Rose Baker.
Using a “dynamic paired comparisons model …utilising barycentric rational interpolation” they have analysed the results of over 20,000 Grand Slam matches involving more than 1,000 players including comparing the performances of some of the all time best tennis players, taking into account their performances from the start to the end of their careers v. the standard of opposition they faced.
Each player’s peak year – the year when they played at their absolute best – has been identified. Every player can be compared with players of any era since 1968 up to the present day. E.g. Jimmy Connors has been compared with Roger Federer based on an analysis of how Connors performed when he was playing against Ivan Lendl, when Lendl placed Becker and when Becker played Sampras and so forth.
Out of the Top 25 All Time Greatest Players eight-time Slam winner Jimmy Connors, at his peak in 1976, is seen as the greatest tennis player ever. Bjorn Borg came second, Roger Federer third and Rafael Nadal fourth. Novak Djokovic, a relative ‘newcomer’ to the game, came 10th.
The statistical model used also shows the ‘golden eras’ of tennis during the mid-1970s, late-1980s to early 1990s and late 2000s by giving the average objective player quality over the period 1968 to 2012. Many will be surprised to see that Pete Sampras does not feature in the top 25 players when measured by peak strength and is at only 15th and 14th place respectively when measured over 10-year and lifetime strength but apparently “he wasn’t beating many other great players apart from Andre Agassi (and)…he often won close matches” i.e. the model is telling us that his Grand Slam wins were at a time when competition was at its weakest.
The research’s findings are definitely provocative … a rankling set of of rankings for some, but above all, it gets people thinking about how we can use numbers to measure and compare performance.
AND we don’t have to wait too long for more…Ian has just completed an analysis that ranks the top English club sides, drawing on every League, FA Cup and League Cup match from 1888. Findings to be published very shortly.