Dr Bill Gerrard, Former Tactical Data Analyst at Saracens and Professor of Business and Sports Analytics, Leeds University Business School
Rugby union and football are both invasion team sports and as such share the common characteristic that the spatial dimension is crucial to understanding why teams succeed or fail. Both sports have experienced similar tactical debates over the relative efficacy of territory-based and possession-based styles of play. These debates have often been characterised as ‘artisan-versus-artist’ with the functionality of direct styles of play to gain territory contrasted against the aesthetics of playing styles in which territory is gained by retaining possession, passing and running with the ball under control. These tactical debates have tended to be associated with differences in attitudes to analytics ever since Wing Commander Reep tried to show that statistical evidence supported the efficacy of the long-ball game in football. To many the statistical support for territory-based play in football provided evidence of the limitations of evidence-based approaches which could not properly capture the quality of possession-based play.
Experience at Saracens
In rugby union the success of Saracens over the last six years seems to parallel developments in football. Saracens, initially under the leadership of the South African international, Brendan Venter, and then the Ulsterman, Mark McCall, have developed a style of play in which the kicking game is a key component. Saracens also embrace the use of data analytics as I know well since I acted as their tactical data analyst from 2010 to 2015. However, the initial adoption of the kicking game pre-dated my involvement by several months and was initially much more an experience-based decision than evidence-based. The statistical analysis supporting the kicking game emerged subsequently.
Between the start of the Venter-McCall regime in the summer of 2009 through to May 2015, Saracens have been very successful, featuring in nine domestic and European semi-finals and five finals, and winning two Premiership titles in 2011/12 and 2014/15.
Implementing the kicking game
Saracens used the kicking game more than any other Premiership team in 2014/15, averaging 20.9 kicks in play in regular-season games. By contrast, the losing finalists, Bath, ranked only 8th with an average of 15.3 kicks in play from their 22 game season.
Graph 1 (below) graphs the relationship between league points and kicks in play for the Premiership 2014/15 regular season for all teams. There is a clear positive correlation between league performance and the frequency of kicks in play even after allowing for the undue influence of the massive outlier, London Welsh, who finished bottom with only one league point and had the lowest game average for kicks in play with only 12.6 kicks in play per game.
The correlation coefficient between league points and kicks in play is 0.546 although this falls to 0.429 if rankings are used to limit the undue influence of outliers. Further evidence of the efficacy of the kicking game is provided by the improved league performance of Exeter in 2014/15. In the previous season, Exeter finished 8th with 45 points and averaged only 14.2 kicks in play, the 2nd lowest in the Premiership that season. In 2014/15, Exeter gained 68 points, the same as Saracens, only missing out on a semi-final berth on points difference. Exeter’s improved performance was closely associated with a much greater use of the kicking game with an average of 19.0 kicks in play, the 2nd highest in the Premiership.
Graph 1: League performance and the kicking game, Aviva Premiership 2014/15
Why so effective?
So why does the kicking game prove so effective in rugby union? Teams that use the kicking game to exit from their own half rather than relying more on ball-in-hand plays minimise the risk of own-half turnovers which are difficult to defend since there is limited time and space to regain a solid defensive shape.
As well as being a high-risk exit strategy, the running game is energy-sapping, particularly with the increased number of breakdowns to be contested (although, of course, this will also impact on the defending team as well).
Tables 1 and 2 provide further evidence of the characteristics of winning performances in the Premiership last season and the relationship with the kicking game. Table 1 uses a win-loss analysis, providing the game averages for selected metrics showing highly significant (at the 1% level) differences between winning and losing performances in 128 regular-season games (excluding four drawn games). As can be seen, winning performances are characterised by:
– More kicks in play
– Fewer rucks won in own half
– Proportionately more play in the opposition half
– Fewer phases of play
– Fewer turnovers and penalties conceded in own half
– Lower error rates in possession
Table 1: Win-loss analysis, selected performance metrics, Aviva Premiership 2014/15 regular season (n = 128 games)
|Performance Metric||Winning Performances (game average)||Losing performances (games average)|
|Kicks in play||17.633||15.406|
|Rucks won, own half||20.922||24.305|
|% rucks won, opp half||69.99%||66.25%|
|Phases per possession||2.658||2.863|
|Penalties conceded, own half||6.148||7.750|
|Turnovers conceded, own half||4.891||6.00|
|Possession error rate||27.73%||33.20%|
Source: Opta data; own calculations
Table 2 reports the correlation coefficients between kicks in play and selected performance metrics across all regular-season team performances. The first point to note is that the kicking game is much more associated with improved defensive performance than with better attacking play. As we would expect, the use of the kicking game tends to lead to fewer points conceded because it makes for more effective exit play and pushes the opposing team deep into their own half, thereby restricting their attacking opportunities. Also, as expected, teams that kick more tend to have fewer phases of play per possession.
The strongest association of any metric with the frequency of kicks in play is the possession error rate (defined as the percentage of own-possessions ending in penalty conceded, kicking error, turnover conceded or scrum awarded to the opposition). The single most important benefit of kicking more is that teams make fewer errors in possession particularly in their own half. Most notably, teams that kick more tend to concede fewer penalties in their own half.
Table 2: Correlation between kicks in play and selected performance metrics, Aviva Premiership 2014/15 regular season (n = 264 team performances)
|Performance Metric||Correlation with kicks in play|
|Phases per possession||-0.267|
|Possession error rate||-0.526|
|Penalties conceded, own half||-0.240|
|Turnovers conceded, own half||-0.050|
Source: Opta data; own calculations
So, in conclusion, there is evidence from the Aviva Premiership last season that teams employing a territory-based playing style with the kicking game used to exit their own half and to force opponents deep do tend to be more effective. Principally it is a risk-averse tactic that minimises error and contributes to greater defensive effectiveness with fewer points conceded, in part because by kicking more teams play less in their own half and therefore tend to concede fewer turnovers and in particular fewer penalties.
But we should note that the kicking game on its own will not guarantee success as can be seen by the respective performances of Newcastle and Bath in the Premiership last season. Newcastle were 3rd only to Saracens and Exeter in the frequency with which they kicked yet they finished 2nd last in the Premiership while Bath, as already noted, ranked below average in the use of the kicking game but finished 2nd to Northampton in the Premiership and reached the final.
From the perspective of the Rugby Union World Cup, it is clear that England are well provided for at fly half to play either a Saracens-type kicking game as directed by Owen Farrell or a Bath-type running game as directed by George Ford.