It’s been awhile since we last posted an article. We used this time to update and optimize our Expected Points (EP) model as well as develop a few new tools that will better assist us in terms of evaluating player and team performance. This will be the first article which makes use of our updated EP model and player evaluation tool. ------------------------------------------------------------------------------------------------------------------------------- Now that the 2021 Six Nations Tournament has concluded, it’s a good time to look back and evaluate the performance of the players. In previous articles we used Expected Points Added (EPA) to evaluate player performances in the 2020 Tri Nations tournament (Article 1, Article 2) and in the discussions within those articles we identified some future work where we could further enhance the model. Here are the performance categories relevant to player evaluation where we made modifications to further enhance our EP/EPA model. This particular set is specific to Wings.
Carry: No changes from previous version Pass: EPA resulting from metres gained directly from passes and offloads were combined into a single category Kick: In the previous version, if possession was kicked to the opposition, the gain in territory is awarded to the kicking player but the loss of possession was ignored. The current version takes into account whether or not a kick was retained by the kicking team. In the current version, if the opposition gains possession as a result of a kick, the kicking player is awarded the EPA for the gain in territory from the kick, but is also docked EPA for the loss of possession.
Try: Tries were evaluated based on historical likelihood of scoring and players were awarded a bonus inversely proportional to how likely the try was to occur. Thus a player that scored a 80 metre individual effort would receive a proportionally larger bonus compared to a player that was a recipient of a “team try” and caught a ball 1 metre from the goal line and just had to fall down to score.
Collections: The previous version only looked at jackals. This category was expanded to include whether or not possession was gained or lost in the following additional situations:
Attacking Catch
Attacking Loose Ball
Charge Down
Defensive Catch
Defensive Loose Ball
In Goal Touchdown
Interception
Mark
Restart Catch
Tap Back
Tackles: Changes in EPA from tackles completed and missed tackles were combined into a single category. The model also took into account the quality of the tackle and rewarded/penalized EPA accordingly. Thus a player who made a dominant, try saving tackle was rewarded with more EPA compared to a player who made a passive tackle which allowed an offload.
Penalties Conceded: No changes from previous version
Turnovers Conceded: No changes from previous version
Here are the results of our analysis of the top performing wings over the entire tournament using the revised Expected Points model.
TABLE 1: Player rankings for Wings in the 2021 Six Nations Tournament based on EPA80 scores. The top 8 players with the highest EPA80 scores that played a minimum of 200 minutes are listed. Each category of performance was normalized to an 80 minute game and, if summed across the table, will result in the EPA80 total.
At first glance, the results appear to pass the eye test. Both Louis Rees-Zammit and Duhan van der Merwe stood out as strong performers throughout this year’s tournament and the fact that both players were named to the Team of the Championship reinforces that the model's ability to measure player performance is in alignment with what we perceive to be strong performances. Where the model provides additional value is in identifying points of difference between players as well as performance in areas that we may not necessarily remember for an entire tournament. In comparing their performances, Rees-Zammit was the better playmaker while van der Merwe was the better tackler but committed turnovers that were more costly. Personally, Monty Ioane was a bit of a surprise, but that is more likely due to the fact that I did not watch any of Italy’s games so I’m not really in a position to comment on his performance. This is the advantage of having a full and complete data set - it fills in the gaps for games that we cannot watch.
It is important to emphasize that these rankings based solely on performance in the 5 games comprising the 2021 Six Nations tournament. This is a very small sample from which to assess performance so I wouldn’t go picking the Lion’s starting back 3 from these results – although there’s no reason why this type of analysis can’t be a tool used to aid selection.
How did your favourite player compare in these rankings? Are there any notable absentees?