Motivation
My desire to gather more descriptive rugby sevens statistics began when I was coaching the University of Michigan. We had been invited to participate in the Collegiate Rugby Championship in June 2014 and had about six months to educate ourselves on sevens after our fall fifteens season. We’d played a few tournaments but had never dedicated significant time to sevens. Although I had strategies and techniques from my time as a player I saw a greater variety in other teams. I was never satisfied blindly allocating team practice time without a decent answer to the question, “Why?”. So I started trying to answer that for sevens.
I’ve spent a fair amount of time playing with advanced sport statistics, mostly in baseball and college football. The dearth of rugby statistics, fifteens and sevens, to this day still surprises. I understand they are both complex, fluid games but there is next to nothing available. Most statistics are taken by individual professional teams with few accepted industry standards. Their statistics are not public for vetting by interested fans. The Aviva Premiership offers some team statistics for their matches but provides little league-wide context. If a team had 212 passes in a game is that a lot or not a lot? Does it matter? If they rucked at 96% for the match should that be celebrated or looked at as a point of improvement? Super Rugby doesn’t even offer these statistics. Where is the baseball-reference.com of rugby?
Baseball is one of the easiest professional sports in which to identify discrete statistics. These stats and their advanced descendants abound online. Their use has revolutionized how people view, coach, and talk about the game. There is still some resistance to relying on statistics to make strategic gameplay decisions. Managers either don’t trust that statistics represent the game accurately or doubt that they were gathered correctly in the first place. Others feel as though the statistics don’t outweigh the human components of the sport. Given the reluctance in baseball, it’s not surprising to find this same sentiment in a very complex non-American sport like rugby where the recording of statistics is much less reliable.
A Cultural Problem of Rugby Stats
Many opponents of rugby stats lean on the human argument; if one player lacks the required courage or cooperation, his or her teammates will suffer. Opponents argue commitment, desire, teamwork, leadership, trust, and heart matter more to a team’s success than what can be quantified on the field. The fact that a player makes ten tackles a game doesn’t matter if they disrupt the team’s culture.
Although I recognize the difficulty such a player would create, coaches at professional levels are paid to basically do one thing, win. If they can win using stats and selfish jerks, they will.
The argument against stats is likely made because rugby gameplay is difficult to quantify as discrete measures. What other game includes muddy situations where a half-dozen players seemingly lay on top of the ball while others yell at them? Dismissing the resource-intensive riddles of rugby’s gameplay protects the coach from uncertainty.
But it is not a matter of either-or. Proponents of statistics aren’t trying to replace clubhouse unity or compassionate teammates. Increased usage of statistics can still allow coaches and captains to develop a cohesive team while providing uniform metrics to measure on-field actions. As seen in other sports, the use of statistics doesn’t undermine the qualities of the sport or how the players feel. The players and coaches still work hard and need to be committed, but now they have more tools for improvement. Personally, I’d want as many tools as possible.
Game State Model Applied To Rugby
An interesting baseball statistic to me is RE24. The basic idea is that you can separate baseball into game states. Bases loaded and zero outs is a game state. Bases empty and one out is a different game state. Over many games, you can quantify the average expected runs each game state is worth. Then as a particular game or season plays out and works through its game states, you can calculate how each team (or player) is doing relative to the expected runs of their game states. In rugby terms, a scrum at halfway is worth some amount of points on average for the possessing team across a large number of games. A scrum five meters from the try line is likely worth more points. A scrum ninety-five meters away is worth fewer. This combination of field position and possession source is your game state.
If a team starts their possession with a scrum ninety-five meters away, earns a penalty, and kicks that to earn a line-out at half-way, this is generally regarded as a positive play and that kicking to touch was the correct decision. But wouldn’t it be great if we knew how many points on average the penalty and kick earned? What if you’re kicking into the wind and only kick to the 22? What if the team only earned a free kick and the other team is going to get the line-out? How far does your kicker have to kick for you to kick to touch rather than take a different option? Rugby’s complexity often reduces this decision to the opinion of the captain or coach at that moment. But the example above has few moving parts and should be easily based on past results. If we had the statistics.
Worse, we often evaluate the decision itself solely by the outcome of the next phase of play. If we lose the lineout after kicking to touch we made a bad decision. If we choose to tap and run the penalty and turn the ball over, the coach throws his notes in exasperated fashion. But that’s not how decision making should be measured. Where the roulette ball lands doesn’t justify the bet. These odds should be known before we make the decision. And that decision should be evaluated separately from how the team performed in the next play.
Methodology
With Google sheets, using sevens games available on YouTube, and student labor, we began recording stats for sevens games to answer these questions. We made modest inroads, recording forty games and likely drawing some hazardous conclusions from the data. Some of it helped shape my view of the game and much of it clarified vague rules of thumb that had been passed down from previous generations. Most importantly, it galvanized my belief that analyzing sevens in this fashion is possible and beneficial.
Though one of our goals for analyzing sevens games was to separate true value from our inherited perceptions of value, we nonetheless were forced to focus our analysis using some amount of personal opinions. The adage with sevens is that it’s all about possession. However, according to what I’ve read on other sports, this shouldn’t be taken as time of possession but number of possessions. The more chances to score, the more points a team will score. Teams are more likely to score more points in soccer or basketball if they get more shots. Holding onto the ball for time may show underlying skill, but it doesn’t get you more shots. Since all rugby possessions either end in a score or a turnover (save those that are purposefully kicked out to end a half), to get more possessions, you need to win kickoffs. Winning the restart battle, thus the possession count, was one of our three main focuses when recording stats.
Next, we looked at the quality of the possessions themselves. I’ve already mentioned finding quality from possession sources and territory. But to measure the quality of open rugby between possession start and stop, we focused on half-breaks and breaks. I believe the offense’s goal is to get behind the defense whether that be running straight through the defensive line, punching through enough to possibly offload, or running around the line. If a team can consistently get behind the defense, they will score.
Finally, once a team is scoring, they need to create quality scores. Which is to say, score seven points instead of five. This has two parts, the quality of your conversion kicker and where you touch the ball down. These are both easily measureable. There’s a reason twenty-one is the “magic number” and not nineteen or seventeen. Converting your tries is not easy and doing so better than your opponents will win you more games.
That is my view of sevens and where we focus our analysis: win the restart battle, create and finish breaks, score seven points instead of five. From a defensive perspective, the inverse is desirable; stop breaks, stop scores (create turnovers), and give up five points not seven. Ruck percentage, pass percentage, tackle percentage, etc. all fall into one of these categories and are more important to How you execute once decisions are made, not Why you made the decision in the first place. I’m not saying these stats don’t provide insight. For example, knowing your scrum and lineout percentages can help you make decisions from penalties. But they are likely most important when individualizing some fundamental knowledge of the game. That fundamental knowledge is what I’m exploring.
Ongoing Work
A few years removed from coaching, I find myself more curious than ever. This site will focus on analyzing rugby sevens using statistics we record. I have expanded what we track in an effort to uncover unexplored avenues of gameplay. Without a team or competition to focus on, I can turn my full attention toward using statistical evidence to reconsider popular assumptions about how sevens should be played. In this project I intend to present interesting and original sevens game analysis focusing on the three main areas mentioned above. As more games are recorded we will build a quantitative foundation for basic sevens strategy. Given the lack of available evidence, I hope what I share is valuable to the broader sevens community. I welcome your input and wish your team all the best on the field.