The Moneyball Approach

The non-selection of the young Stormers loose-forward, Siyamthanda Kolisi, has sparked somewhat of a debate around South Africa, as many felt that he was a better player than the incumbent Springbok openside flanker, Marcell Coetzee. Whilst reading through some of the comments made on various social media platforms, I found myself fairly ambivalent on the issue as no one could provide me with a convincing argument. Seeking clarity, I took it upon myself to delve deeper into the issue. 

It was 2002 and Billy Beane, the General Manager of the Oakland Athletics, a Major League Baseball Team, was faced with a dilemma. The Athletics had narrowly lost in the play-offs to the Yankees in the previous season and were now facing the prospect of losing their three star players to free agency. Facing budgetary constraints which seemed insurmountable, the Athletics set about replacing those lost stars. 

Convention was thrown out the window as the Athletics embarked on a campaign of adopting a revolutionary approach to the game using ‘Sabermetrics’- the specialized analysis of baseball through objective evidence- which became popularly known as ‘MoneyBall’. The basic idea was to replicate these lost stars in aggregate as opposed to simply replacing them man for man.  In order to do so, they had to sign seemingly average players, at a discount, who were undervalued by the system that coaches and managers were currently using. The decision to sign these particular players was not based on their potential or because they had an attractive swing but rather using a statistical approach based on the player's ability to get to first base. Once signed, the players were guided by statistics on which shots to play for them to generate runs. As the process became more accurate over time and by defying all existing baseball logic, the Athletics were able to develop a winning formula which resulted in a twenty game winning streak during the season.

The approach is fairly simple, indentify a weakness and mitigate it. Identify a strength and amplify it. Although significantly different from Baseball in terms of variability, the game of Rugby can learn a great deal from this approach. Player stats are easily attainable but when they are viewed in isolation they don’t really mean much. The difficult part is finding meaning in the stats which can then be applied in developing a winning model. 

The Kolisi-Coetzee conundrum should be no different. Both of these talented youngsters played a significant role this season as ball carriers and defensive stalwarts for the Stormers and Sharks, respectively.

Using the Moneyball Approach and considering the game plan currently being employed by Heyneke Meyer, which player was better suited for higher honours?

Super Rugby 2012 stats. (Courtesy of Fox Sports)

Coetzee

Minutes played: 1333

Ball Carries: 195

Tackles Bust: 53

Offloads: 26

Linebreaks: 5

Tackles Made: 273

Tackles Missed: 36

Kolisi

Minutes played: 1167

Ball Carries: 81

Tackles Bust: 31

Offloads: 5

Linebreaks: 3

Tackles Made: 223

Tackles Missed: 22

On face value these stats tell you that Coetzee does more on the park than Kolisi and he may, therefore, be assumed to be a better player. With a slight bit of manipulation, take a look at this:

Coetzee:

Carries per Tackle bust: 3.68

Carries per line break: 39

Carries per Offload: 7.5

Tackles per Missed Tackle: 7.58

Kolisi

Carries per Tackle bust: 2.61

Carries per line break: 27

Carries per Offload: 16.2

Tackles per Missed Tackle: 10.13

What the stats now tell you is that Kolisi is a more effective ball carrier and defender. The approach, considering the Springbok game plan, should’ve been to select Kolisi and give him the ball more often. At the Stormers, Kolisi is subjected to a game plan which is focused primarily on defense and, as a result, does not get the opportunity to carry the ball as much as Coetzee but when he does he is far more effective. The mere fact that he is playing at a franchise which is not attack minded should not be the deciding factor on whether or not he becomes a Springbok. 

This type of approach to selection would ensure that the best player gets picked based on facts and not on the gut feel of the selectors. If the game plan was one of continuity then the selection of Coetzee would make sense considering his superior offloading skills, but how many times has he offloaded the ball for Heyneke Meyer's Springboks this season? This statistical approach does not stop at selection. What should then happen is an analysis of Kolisi’s ball carrying in terms of where on the field the majority of his tackle breaks come from, e.g. how far away from the ruck is he when he gets the ball and breaks a tackle? Once these stats are available, it would then make sense to put Kolisi in these positions more often to amplify his ball carrying ability. This approach should not come across as being ground breaking, it just makes logical sense.

Although skills development should be the long term approach for Springbok success, I feel that selecting the right players in the right positions for the desired game plan is of paramount importance in the short term. For instance, it wouldn’t make sense to select Mvovo if you plan to kick constantly on his wing because he is not the best player in the air. If you plan to run the ball, however, his selection begins to make sense. We need to start looking at these decisions not based on emotion but rather on cold, hard facts. If a player is performing and is best suited for a particular game plan then he needs to be selected.

At international level, coaches are lucky enough to decide on a game plan and then select the appropriate players. At club and university level, the available players often dictate the game plan. Gaining an understanding of this dynamic will lead coaches to develop a winning strategy which best suits their circumstances. I feel that the use of statistics and video analysis will aid them in doing so. The Moneyball approach is yet to influence Rugby as it has Baseball, but with continued experimentation and refinement we could potentially see average teams transform into winning ones.