Skip to Main Content

Team performance analysis in rugby union

By: Stats Perform

This is the second in Neil Watson’s series of blogs for OptaPro. In these articles Neil approaches data analysis within rugby union from an academic standpoint while maintaining a focus on real-world applications.

Neil is a Statistical Sciences lecturer at the University of Cape Town.


This is the second post in a series of blogs that will consider various aspects of team performance. In the previous post, I considered 66 Key Performance Indicators (KPIs) that had previously differentiated winning teams from losing teams. I answered two questions: (i) Which KPIs are still valid differentiators? And (ii) Which KPIs differentiate across different competitions?

In this blog, I consider only the top and bottom teams in the European Champions Cup (ECC), Heineken Cup (HC) and Super Rugby (SR)[i]. I chose the two finalists and two teams having the lowest log points in each tournament and ran the same analysis as in the first blog. A part of the results is shown in Table 1 below.

Table 1. Distribution of statistically significant KPIs

A few highlights (with comparison to results from the first blog):

i. A big increase in the proportion of KPIs displaying medium effects (17.6% to 6.9%) and a much smaller proportion displaying negligible effects (11.8% to 22.4%).

ii. As before, the KPIs showing the largest effects all involve those variables strongly correlated with scoring points. The top teams score more than twice as many points, tries, and conversions.

iii. A larger difference in the % possession between winning and losing teams (8.2% to 5.98%).

iv. When examining the significant KPIs across all three competitions, top and bottom teams differ most in the areas of attack and territory.

Two things to consider when interpreting the results above: Firstly, those KPIs showing large effects are strongly correlated with one another. Secondly, one should be mindful that most KPIs here are counts of actions, instead of ratios (%) that bring context to the performance of a team. This can result in their interpretations being misleading. For example, if one examines Rucks and pass and Rucks lost, top teams lose fewer rucks and also pass directly from the base of the ruck less frequently. But unless we have information on the average total number of rucks per game, we cannot make a definitive interpretation. Perhaps the bottom teams have a higher number of rucks per game? The issue of expressing variables as counts rather than proportions/ratios is important and something I will discuss in a future blog.

To facilitate the comparison of top and bottom teams, I developed a rankings matrix of 12 KPIs of interest across six areas of play – attack, defence, discipline/errors, kicking, possession/territory and set piece. Table 2 displays the rankings for the top two and bottom two teams within each competition. For most KPIs a higher ranking is better, but there are a few where a lower ranking is better (% of tackles missed, Penalties conceded, Errors and Lineouts lost).

Table 2. Team rankings in a selection of KPIs across areas of play (Teams: SRC = Saracens, TLN = Toulon, TRV = Benetton Treviso, ZBR = Zebre, CLR = Clermont, CST = Castres Olympique, SS = Sale Sharks, HGH = Highlanders, HRR = Hurricanes, BLS = Blues, FRC = Force)

A number of results warrant discussion here, but I will include only a couple of thoughts for each aspect of play:

Attack: Top teams rank far higher for the number of Linebreaks per game, indicating that they often found ways to break their opposition’s defensive lines. It would be intriguing to further investigate the strategies utilised to achieve this – for example, is the aim to put players into space or create mismatches between backs and forwards? The difference in Points scored when possession starts in opp22m ranking is smaller, but we can infer that top teams are more clinical in converting scoring opportunities in the opposition’s 22m.

Defence: Top teams’ defensive systems are generally better at slowing the momentum of the opposition (with a few exceptions). However, it is also evident that having a good defence isn’t necessarily a guarantor of success. For example, Toulon (TLN) didn’t have a great defensive season in the HC yet they won. They clearly placed appreciable emphasis on this area in the next season, where they ranked much better on these KPIs. Another example is the Sale Sharks who achieved good rankings on these KPIs but finished with the lowest log points.

Discipline/errors: In both the ECC and HC, top teams generally conceded fewer penalties and made fewer errors. SR is interesting in that the top and bottom teams are closely ranked. In fact, both the Highlanders and Hurricanes made more errors than the Blues and Force across the season. Does this indicate that the two finalists played with greater freedom in that they weren’t focused on minimising errors? Further analysis of additional KPIs will help answer that question.

Kicking: Top teams kick more and/or recover a higher number of their kicks than bottom teams in general. A notable exception to this Toulon in the HC, where they ranked poorly on both KPIs. Again, they appear to have examined this area with intent as they improved in both in the ECC.

Possession/territory: Top teams enjoy a greater % of possession and enter the opposition’s 22m more frequently. The rankings here reveal that top teams hold possession of the ball for longer periods and have more effective tactical kicking games.

Set piece: In general, top teams lose fewer lineouts. A notable exception here is the Hurricanes, who lost the most lineouts of any SR team. I’m intrigued as to how much they improved in this area this year as they are the current SR champions. In rankings for Mauls won there is less distinction between teams within competitions. I think this points to there being few mauls per game, and the difference between the top and bottom teams is not large (5.78 to 4.06).

What next?

A question of great debate in the global rugby community is whether specific game plans exist that contribute towards winning. The next blog will compare top vs bottom teams in greater detail, to contrast the different game plans adopted within each competition and across competitions.

[i] There was not enough data on the reduced amount of teams in either the Six Nations tournament or the Rugby Championship to conduct any meaningful analysis.