Own up. Have you ever ambiguously praised a player’s ‘unseen’ breakdown work or noted how someone seems to knit together a team performance without being able to pinpoint any eye-catching moments?
Interception tries are sometimes said to have caused a ‘14-point swing’. Perhaps you have pondered the exact value of a side-step or a lineout steal. Now, such truisms and bold statements will be properly interrogated. Vague terms like ‘game management’, which have established themselves in rugby union’s lexicon, can be bolstered by substance. And science.
A data revolution has been gathering momentum quietly in the 15-a-side game for some time, very occasionally peeping out of the shadows to show itself. This weekend, after years of painstaking development, one sophisticated and exciting project will step into the light for public consumption. Once there, it could overhaul and enhance the way in which many supporters watch the sport.
As the United Rugby Championship (URC) gets under way, television viewers will be exposed to the Expected Points (xP) metric – rugby’s answer to Expected Goals (xG) in football and the Expected Points Added (EPA) ratings of the NFL.
During URC matches, featuring sides from Ireland, Italy, Scotland, South Africa and Wales, every meaningful action on the field will have a points value attributed to it in real-time. Heavy tackles, offloads, penalties and plenty more will be weighted according to estimated impact on the final score and totted up accordingly. Prepare to gather debate-settling ammunition… or to have preconceptions shattered.
The xP system has been devised and rolled out by James Tozer and Gordon Hamilton-Fairley, who co-founded Oval Insights last year. Both have been name-checked by Eddie Jones as influential figures in England’s turnaround following a grisly 2018.
Tozer calls on classic board-game imagery to explain how the software assesses a player's actions.
“Rather than looking solely at the player who touched the ball down for a try or kicked it through the posts, we estimate the impact that every action has on a team’s chance of scoring,” he continues. “Imagine it like a game of snakes and ladders. There are some actions that move you closer to scoring (the ladders) and some that move you further away (the snakes).
“The difficult bit is working out how long the snakes and ladders should be – how beneficial is a line break, and how detrimental is a handling error? A standard approach in rugby has been to come up with subjective rating systems, based on gut feel. What makes xP different is that it is objective: a combination of statistical models derives the value of each action, rather than us.
“For example, a line break increases a team’s probability of scoring, leading to an average benefit of +0.3 points on the scoreboard. A passing error has the opposite effect, of -0.4 points. Rather than awarding a try-scorer 5 points, you can split them among all the teammates who helped (or hindered) along the way.
“One very clear and important insight from xP is how costly errors can be. Spectators tend to remember occasional bits of individual brilliance, but those can be undone by lots of small mistakes. Everyone loves watching Leone Nakarawa make outrageous offloads, for example, but last season [for Glasgow Warriors] he ended up on negative xP because of how often he gave the ball away.
“Conversely, winning the ball back is really valuable. Jac Morgan, an uncapped 21-year-old Welsh flanker [formerly of Scarlets, now with Ospreys], is near the top of the xP table on URC Statmaster [the competition’s in-house ranking] because of how often he won turnovers at the ruck.”
Tozer and Hamilton-Fairley say a lack of historic investment in data science has left rugby significantly behind American football, basketball, baseball and football when it comes to answering one simple question: does a player’s action on the pitch help their team to win?
“When people have tried to make rugby more sophisticated, they have generally done so by counting more events or combining them into new metrics, rather than statistically modelling their impact on scoring points and winning games,” Hamilton-Fairley adds.
Oval Insights is well down the line when it comes to harnessing the potential of its intellectual property. It can assess performance – coaches may be encouraged when their team’s xP is higher than their actual score – and recruitment is another area that could be influenced by xP, which obviously becomes more interesting when one extrapolates it over a season or over a player’s career.
That said, Tozer and Hamilton-Fairley are avid rugby fans. They appreciate – and, indeed, relish – rugby’s intangible aspects such as 50-50 refereeing decisions or character references when it comes to signing players. They also suspect that xP will be met with cynicism in some quarters.
“We know from working on the performance side that great players do lots of intangible things, and that match context is really important,” Tozer finishes. “We make lots of additional adjustments for those sorts of things when analysing selection or recruitment.
“But the overall aim with this public-facing version of xP is to give fans a more advanced understanding of the sport they love. That’s why we’re taking the extra step of publishing some of the underlying xP values, rather than just the final results.
“All we ask is that fans keep an open mind: if someone scores higher or lower than you expect, it may be because they are doing lots of crucial stuff that is easily forgotten after the final whistle.”
Expected points in action
Values for the build-up to Michele Lamaro's try for Benetton against Vodacom Bulls in last season's Guinness Pro14 final.
Paolo Garbisi kick retained: +0.40
Monty Ioane line break: +0.30
Monty Ioane carry (10 metres): +0.10
Monty Ioane pass assist: +0.30
Michele Lamaro defender beaten: +0.10
Michele Lamaro carry (11 metres): +0.11
Michele Lamaro try: +1.40