Congratulations to the Exeter Chiefs for winning the 2020 Gallagher Premiership Championship by defeating Wasps by a final score of 19 – 13. Following this match there was a lot of spirited discussion on social media about the Wasps’ decision to kick to the corner for a lineout versus taking the points from a penalty kick. For those of you who did not get to watch the game, the Wasps were down 16-13 to Exeter at the 73:15 mark and were awarded a lineout 5m away from the Exeter try line. Wasps won the lineout and set up a driving maul which resulted in Exeter being penalized for pulling down the maul. Ultimately Wasps backed themselves and elected to kick to the corner a second time and take their chances on scoring the try. This decision was made knowing that the weather conditions were deplorable and Wasps had a 22 year old replacement hooker to throw into the lineout – so not without its risks. Regardless we can appreciate that the decision was aggressive and made with the mindset of winning a championship.
This was a pivotal moment in the game. There were no guarantees that Wasps would have converted their penalty kick and, in doing so, would have only tied the game with ~4 mins left in the game. As hindsight is 20/20, we know that the penalty kick wouldn’t have mattered as Exeter took possession of the ball and kicked a penalty goal of their own on the last play of the game. As it is possible to have a good process but a bad outcome, does the fact that Wasps turned the ball over on their own lineout throw mean that kicking to the corner was the wrong decision?
For all of you rugby nerds, to better help us evaluate the decisions made, we have developed a Win Probability (WP) model which was assembled by analyzing over 500,000+ game actions which took place over 9 seasons across all major rugby competitions (e.g. Gallagher/Aviva Premiership, Super Rugby, Guinness Pro 14, Top 14, International test matches, The Rugby Championship, etc.). This WP model did not take weather conditions into account, but we are currently looking to incorporate this into future versions of this model. So the insights gained from this model are for “average” conditions but for this analysis we will also consider and discuss the potential impact of the weather.
Given the specific game states for that important lineout (e.g. time remaining, field position, score differential) and given that a 5m lineout in the Gallagher Premiership is worth 1.95 EP, by electing to go with a lineout Wasps had a 41.1% chance of winning the game. If Wasps were successful in kicking the penalty goal, they would have had a 48.9% chance of winning the game, however this had to be balanced with the possibility that they could also miss the kick. If they missed the kick, Wasps’ chances of winning would drop to 27.7% due to not scoring the points as well as the loss in territory resulting from receiving the associated 22m Restart kick (NOTE: the average starting position from a 22m restart is 42 metres from the opposition goal line).
The break-even conversion probability (x) for the decision to make it worthwhile to kick the penalty goal can be found by setting the value of the lineout equal to the total value of the penalty goal attempt:
Thus, if Jimmy Gopperth had a better than 63.9% chance of making the kick, Wasps should have taken the points from the penalty goal. Given that Jimmy Gopperth kicked penalties/conversions at an 86.8% success rate in the 2019-2020 Gallagher Premiership campaign, the decision should have been a slam dunk – Wasps should have kicked the penalty.
Well…..no. This is where a superficial analysis of the "stats" and lack of context can be costly. While it is true that Gopperth is a very accurate kicker, if we look at a scatter plot of where he made/missed his kicks over the past 7 seasons (2014-2020), we see a very interesting trend (Figure 1). The circled area of the figure identifies where Gopperth would likely have attempted the kick. It is clear that this is an area of the pitch where he struggled and, in fact, he only made 56.0% of the kicks attempted. Thus the goal kick would have been worth ~1.68 EP (3 x 0.56 = 1.68). Now you also layer in the poor weather into the decision and it becomes evident that attempting the penalty goal is less than optimal.
FIGURE 1: Summary of all penalty goals/try converts attempted by Jimmy Gopperth in the Guinness Pro 14 and Gallagher/Aviva Premiership over the past 7 seasons
(2014-2020)
Taking all of this into account, it appears that attempting to score from the lineout was the better option. However, we know Wasps took this option and still lost the championship – does this mean that they were doomed to lose regardless of whether they chose lineout versus penalty goal? The data does suggest this since both options result in win probabilities of less than 50%.
Well all is not lost. Even though we made the case that the lineout was likely the better option compared to attempting the penalty goal, it doesn’t mean that it was the best option available. Remember that Wasps had other options – they could have taken a scrum or a tap penalty. In fact, if we compare the expected points of all opportunities from 5m out it’s clear that there was a better option (Figure 2).
FIGURE 2: Expected points for lineouts, scrums, and tap penalties when attacking team is 5 metres away from the opposition goal line in the Gallagher/Aviva Premiership
(2013-2020)
The tap penalty was the best option for this scenario with an EP of 3.47 which is statistically greater than either the lineout or scrum. Knowing this, you don’t even need to think about kicking the penalty as the expected points are greater than the value of converted penalty kick. Had Wasps elected to go with the tap penalty their win probability would have increased to 53.6%. In addition, the tap penalty was arguably a safer option and was more likely to be executed without error given the challenging weather and substitute players in the game. Ironically, Exeter have drawn attention to this strategy with their recent success from taking tap penalties in the red zone. Article 1 Article 2
While it is highly unlikely that players will have memorized these numbers and would have been able to recall it under game pressure at a critical moment, this is something that can be discussed in the post facto game review to educate players about future decisions in similar situations. At this point it is not clear which of 1) “quick tap” penalties or, 2) tap penalties where teams take time to set up prior to the tap, is the better tap penalty strategy. Nonetheless, this is an area where data analysts can collaborate with video analysts to come up with an answer. As a word of caution, these insights around the tap penalty are specific to the Gallagher Premiership and may not necessarily translate to other competitions (professional and/or amateur).
This is an example of where advanced analytics can be used to shape decision making for rugby teams. Inevitably there will be staunch traditionalists that will resist insights found in big data, but this is great for initiating further discussions about the game. If you were in Joe Launchbury’s shoes making this critical decision in a championship game, would you have called for the tap penalty? Now that we know that tap penalties are the option that yield the greatest expected points 5m from the opposition try line, do you think teams will start to use this strategy more in the red zone?