Here is the second part of a series looking at what the advanced analytics say about players Part 1, which looked at the backs, can be found here. The same criteria for inclusion in the study were also used for the forwards. For this study, we looked at the following actions for forwards (normalized to 80 minutes):
Carry metres
Kick metres
Pass metres
Lineout receive
Lineout throw
Scrums
Jackals won
Tackles completed
Missed Tackle Impact
Penalties conceded
Turnovers conceded
What stood out to me in this analysis was how setpiece influenced the scores of the tight 5 players. In general, the players with the highest scores were typically amongst the leaders in the setpiece metrics (lineout/scrum) within their positional cohorts. More specifically, the numbers favoured players that played on teams that play a style of game which focuses on setpiece. For example, Adriaan Strauss and Bismarck DuPlessis had very high lineout throw scores because the Springboks lineout was arguably the strongest in the world at the time, thus they played a style of game which would maximize the number of lineouts they could run in a game. Thus, Strauss and Bismarck benefitted from the fact that South Africa would run more lineouts than other nations. While this analysis represents the actual numbers derived from the data, some coaches may not agree that a player’s value cannot be assessed strictly on performance in setpiece. We will discuss how this effect can be mitigated later in the article.
One of the obvious limitations of this analysis is that outside of the jackal – the breakdown was not included. While breakdown data was avalable, it was not included in the study because it did not exist for all competitions over the period of the study – thus it was incomplete and would have favoured a certain subset of players over the others. One could speculate that the standing of several players in this analysis was impacted by the omission of contributions to the breakdown, but this is something that we will look to remedy in future analyses.
Another limitation of the analysis, is that only specific players were tagged in their contributions to setpiece:
Lineouts – only the thrower and receiver were tagged
Scrums – only the props were tagged
For lineouts, the thrower and the receiver shared the EPA allocation for lineouts – each received 50% of the value. Other players participating in the lineout (e.g. lifters) were not identified in the data and therefore could not be allocated of the portion of the EPA. For scrums, the props were credited with the full EPA value of the scrum. There will be some of you who will argue that they only constitute 25% of the composition of the scrum (2 out of 8 players) so the EPA should be allocated accordingly (NOTE: there are other ways of allocating contribution to the scrum – this is outside the scope of this post and will be discussed in future discussions around the scrum). While this part of the player evaluation model isn’t perfect, I’m sure there are some front row players who are thinking “damn right we should get credit for the scrums, especially since we don’t get credit for lifting in the lineout!”.
Now that we have the formalities out of the way, here are the rankings:
NO. 8 (Table 1)
While there is no question that Sergio Parisse has been a talisman for Italian rugby, the numbers suggest that Kieran Read might be the top performer in this cohort. What is most impressive is that he played significantly more minutes than everyone else and was still able to consistently contribute at a high level.
TABLE 1: No. 8 player ratings (EPA_80) based on performances in the Rugby World Cup, Six Nations, and The Rugby Championship. World Rugby Team of the Decade selections are highlighted in yellow.
BACK ROW/FLANKER (Table 2)
The data identified back row (#6) and flanker (#7) as separate categories even though many of the players in this list have played both. For this reason, this group was amalgamated into a single category. Richie McCaw and David Pocock were both selected to the Team of the Decade – do you think their EPA_80 scores would change if the breakdown data was included? Some of you may point out that Richie McCaw was voted Player of the Decade and that these numbers do not support his selection. My rebuttal to that is that McCaw retired in 2015, so these numbers reflect his performance in the last 2 years of his career and do not capture him in his prime. Context matters.
TABLE 2: Back Row/Flanker player ratings (EPA_80) based on performances in the Rugby World Cup, Six Nations, and The Rugby Championship. World Rugby Team of the Decade selections are highlighted in yellow.
LOCKS (Table 3)
A lot of really strong players in this cohort. But there does appear to be some bias at the top towards teams that emphasize lineouts in the game plan.
TABLE 3: Lock player ratings (EPA_80) based on performances in the Rugby World Cup, Six Nations, and The Rugby Championship. World Rugby Team of the Decade selections are highlighted in yellow.
HOOKER (Table 4)
Just like the locks, the numbers favour players that throw in a lot of lineouts.
TABLE 4: Hooker player ratings (EPA_80) based on performances in the Rugby World Cup, Six Nations, and The Rugby Championship. World Rugby Team of the Decade selections are highlighted in yellow.
PROP (Table 5)
Performance in the scrum was based on the following formula which aggregated all scrum outcome EPAs from a team’s own put in (own) and the opposition put in (opp).
SCRUM_80 = EPA Scrums/Pens won (own) - EPA Scrums/Pens lost (own) + EPA Scrums/Pens won (opp)
It is important to note that penalties conceded did not include scrum penalties. The reason for this is that penalties conceded were already accounted for in the scrum evaluation, thus we did not want to double penalize props in their overall evaluation.
TABLE 5: Prop player ratings (EPA_80) based on performances in the Rugby World Cup, Six Nations, and The Rugby Championship. World Rugby Team of the Decade selections are highlighted in yellow.
THE HYBRID PLAYERS (Table 6)
Once again there were players who played multiple positions over the course of the study. Most of these players are players who started in the back row and made the transition to the second row. Eddie Jones brought attention to this type of player in England’s build up and participation in 2019 Rugby World Cup and this category may grow in future versions of this analysis.
TABLE 6: EPA_80 player ratings for notable hybrid players based on performances in the Rugby World Cup, Six Nations, and The Rugby Championship. World Rugby Team of the Decade selections are highlighted in yellow.
FUTURE WORK
Earlier in this post we discussed how a team’s playing style can influence a player’s EPA_80 score. Specifically, players that play on teams which play a setpiece dominant style will have higher lineout/scrum scores than players from teams that opt to play a more “free-flowing” game. If I’m a coach that prefers to play a looser game, I may want player evaluations to reflect that. One way to correct for differences in style of play is to change the weightings of the measures so that the desired aspects of performance are more heavily weighted. The risk of doing this is that you may distort the data in such a way that it becomes less effective with regards to measuring team and individual performances. In the future, if there’s enough interest we can look at optimizing the EPA_80 metric by playing with different weightings.
How did your favourite forward perform in this analysis?