This post will outline some of the main statistics you’ll see on this site. In particular, I’m attempting to provide clarity with regards to possessing the ball. In my opinion, rugby possession isn’t as clearly defined as in many other sports and the possession sources vary greatly. These different sources yield different expectations of scoring and, as such, should be evaluated differently. Some statistics that I use require detailed explanation. Some of these stats I devised myself and they are still developing as I think about and apply them.
Possession is a common term but I don’t think the definition is uniform within the rugby community. For me, a possession occurs when the ball is in a team’s hands in open play. Setting aside the rare penalty kicks, you can only score if you have the ball in hand. I do not consider lineouts, scrums, or restarts to be possessions until the ball is actually taken in hand by a team.
Teams may also advance the ball through kicks in live play. Sometimes this occurs directly from a turnover before the ball is taken in hand. I don’t consider this a possession until they have picked up the ball. In this way, kicks from hand can be seen as a momentary lapse of possession. I won’t tally a second possession if they regain the kick, but I feel it’s important to clarify when possession does and doesn’t occur. Further, loose passes, scrums, and lineouts are other examples of when possession is in doubt.
Scrums, lineouts, restarts, turnovers, kicks, and penalties are sources of possessions. Though they don’t necessarily provide the team in control of the ball with possession.
Because retaining possession at a kick, restart, or set piece is in doubt, I will call these “opportunities”. Opportunities are situations where a team is expected to get the ball in their hands. Having the throw-in at a lineout or the put-in at a scrum or being set to receive a restart are examples of opportunities. The team in control is expected to gain possession, but the other team may gain possession first. These occur more often than possessions since all possessions come from an opportunity but not all opportunities result in possession.
Though I’ll often reference the count of opportunities for teams, different opportunities come with different chances of possession and thus, should be valued differently. Teams win their own scrum around 95% of the time while they only win their lineouts around 78% of the time. Both are opportunities but they don’t offer the team in control of the ball the same chance of obtaining possession. These situations are opportunities for their opposition too. Though their expectation of gaining possession is much smaller, well below 50%, so it’s best to view the opportunity from the perspective of the team in control of the ball (but not in possession!).
I currently count opportunities at scrums, lineouts, restarts, turnovers, penalties, and kicks when a team is able to kick the ball before “possessing” the ball. But I plan on using the relative likelihood of gaining possession from each type of opportunity to provide a more precise “opportunity” value. All the while, we can use the chance of retaining possession and the chance of scoring from this possession to calculate an expected number of points for teams across all of their opportunities.
Sequences are the sections of live rugby that unfold between whistles and turnovers. A given team’s “possession” may have more than one sequence if the defense commits a penalty or knocks the ball on for example. Team A may win a scrum, play some rugby, win a penalty, kick to touch, win a lineout, play some more rugby and score. That was really three sequences — scrum to penalty, kick to touch, lineout to score.
We will call this a possession though Team A didn’t truly obtain possession until they won the scrum. Before that scrum win, they merely had an opportunity, and this particular opportunity was worth .95 possessions.
This model of opportunities, possessions, and sequences has the virtue of making the game look extremely simple. It divides the game into a series of opportunities, passed back and forth, between the two teams. These opportunities yield possessions which have three outcomes: teams can score, turn the ball over, or end the half. These outcomes are easily evaluated which allows us to evaluate the possession as a whole.
I often mention “expected points” in my posts. This is the average number of points that teams in the same situation have scored. The situation is given in terms of location on the field and game state (lineout, ruck, loose ball, etc.). When evaluating opportunities, one needs to factor in the chance of first obtaining possession. Evaluating across sequences requires accounting for how a team has progressed without turning over the ball. As mentioned above, teams may advance the ball and move to a different game state but still have possession of the ball. At that point, we can compare the expected points of the new situation in comparison to their starting situation and determine whether the team is in a better or worse position to score.
Also, consideration needs to be given to what happens if you turn the ball over. There is a small expectation of scoring from your own tryline. But a turnover here is very costly as the other team is very likely to score. So while possession at your own tryline has a positive rate at which teams score, considering the likelihood of a turnover and the relative harm of a turnover, the total expected value of the possession is actually negative. This is why many teams kick the ball from their own tryline.
22 restarts don’t follow the typical model of opportunities moving back and forth between teams. If Team A kicks the ball ahead but it goes out of the back of the try zone, they receive a 22 dropout. Team B gets to hold that ball and kick the 22 where they want, but it’s not in live play so they don’t have possession. This is really another opportunity for Team A as they are set to receive the ball. Receiving a second chance on your possession is significant when teams receive around eight opportunities a game.
Alternatively, Team A’s kick into the try zone could have been picked up by Team B. Team B could play from their own try zone and possibly run the ball out. But more than likely, Team B grounds the ball and kicks a 22. Team B had possession of the ball but isn’t expected to attack, let alone score. This possession needs to be accounted and not analyzed the same as a possession where Team B was truly trying to score. This is very similar to possessions where a team kicks the ball out to end a half. I will exclude these possessions from the majority of analysis.
This is meant to further clarify some terms I use to describe the game of sevens. Opposing teams usually swap opportunities back and forth but possession isn’t always exchanged. Turnovers and whistles chop the game into sequences and provide clear game states with which to measure how teams are progressing through their possessions. Though the evaluation of live play will be more difficult, a clear framework of dead ball game states goes a long way to evaluating teams beyond the final score.