Rugby's data revolution - can Moneyball be oval?

"I've watched more rugby in the last five years than I think I'll ever watch in the rest of my life."

And for Angus Campbell, that really is a lot of rugby.

On a weekend, Campbell will gather with his Oval Insights colleagues in an office on an industrial park outside Leeds and watch live feeds of games happening simultaneously around the world.

It sounds like a sports bar. It is, in fact, hard graft, as they harvest data at scale.

For each match, two people work furiously to categorise the main elements of the action as it unfolds, marking possessions, penalties, set-piece outcomes and open-play kicking.

Another four people follow behind, taking passages of play in turn, and enriching that initial information with finer detail.

Individual carries, post-contact metres, tackle success, tackle type, tackle dominance, individual ruck entries and ruck duration are just some of the information they are hunting.

Shortly after the final whistle, they will have broken each match down into 2,000 individual events.

With the various descriptors attached to each, there is a total of about 10,000 pieces of information recorded for every match in each significant top-tier tournament.

It is a mountain of data, which is funnelled into Oval's Advantage database.

All 10 Prem sides, just over half the United Rugby Championship (URC) teams and a significant number of Top 14 clubs subscribe to get access.

Product manager Campbell, along with Oval's customer success manager Sam Kisbee, helps clubs make the most of what is inside.

In the 'Moneyball' era, where data has transformed recruitment in baseball, football and beyond by unearthing overlooked talent, is it now doing the same for rugby?

Alex Sanderson oversaw one of Prem's biggest squad turnovers this summer, with 12 new players arriving at Sale.

"When you have got a raft of possible players, stats are always a big factor to help understand which of them fit your game model," he says.

"We know how many metres we make post contact, what our defence depends on, and what attacking decisions we make and you can map those metrics.

"You can create a spider graph to find players that fit the 'shape' that you want for a particular position."

Courtney Lawes was Sale's headline signing. The numbers he was registering in ProD2 with Brive, helped convince Sanderson that the former England captain was worth a one-year deal.

In a similar way, stats also helped Phil Dowson, Sanderson's counterpart at Northampton, work out how to replace Lawes when he left Saints in 2024.

"We also have a game model around what we want different positions to look like," he said.

"If you can't get everything you need in a position, at a price point you can afford, you have to make amends elsewhere.

"So to replace Courtney Lawes, we needed some line-out ability, some turnover ability, some ball-carrying and also some leadership.

"The game is now being coded in so much more depth than previously and the challenge is to use that to separate the wheat from the chaff.

"We try and find people who fit the bill, trim the list depending on what we can afford and then do due diligence by looking more deeply at their performances."

The edge for clubs now is no longer the data itself, but the ability to identify what is most important amongst it all.

Oval has developed a model which flags up players who have outperformed its expectations in different aspects of the game.

But clubs are also looking externally for expertise.

Sale are bringing on board an artificial intelligence company, recently bought by billionaire owner Simon Orange.

It will initially crunch numbers on injury prevention and player availability, but Sanderson plans to add recruitment to its tasklist.

Northampton have had help from the world of Formula 1.

Ayao Komatsu, the Haas team principal, was a rugby player and is still a rugby fan. The 50-year-old turned out for Old Wheatleyans and Loughborough when he moved from his native Japan to study in the Midlands in the 1990s.

He has come into Northampton to talk over his approach to data with Dowson.

"For Ayao, the most important thing is the driver," says Dowson. "How does he feel? For Haas, the driver's backside is the best receptor of data.

"Even in a sport like F1, which is so data driven, he is still going to the human element of it. That is something that resonates with me."

Sanderson also emphasises that while data can sift and shortlist possibilities, there is a human touch to recruitment.

A coach's eye can pick up on details around a carry or clearout which might be missed by a blunt database definition.

Just as importantly, in a sport in which success often hinges on cohesion and unity, any new arrival needs to fit in as a person, as well as a player.

"That data dive and performance analysis will only get them to the door, really," says Sanderson.

"Then it's all about character. It has to be about character, because there's quite a lot of players that consistently look quite good, but won't be the right fit for the group."

Oval Insights has grown from co-founders Gordon Fairley and James Tozer's two-man band.

Six years ago, the pair would carry out bespoke projects for clients, coding games themselves from their flat in Balham and writing reports.

As their operation has expanded, they way Oval works has also changed. Given their wide customer base, identifying specific players would present potential conflicts of interest.

"We have to now say 'we'll give you the tools and whoever uses our tools the best is going to find the gold the quickest'," says Campbell.

"Ever so often though you will see a player move and we will say to each other, 'that's a good signing'. Or, alternatively, that it isn't!"

Despite the advances, rugby isn't at the forefront of sport's data revolution.

Some of that is down to the nature of the action.

Baseball, where the term 'Moneyball' was born with the Oakland A's at the turn of the millennium, consists of a collection of players' individual actions, usually not in conjunction with their team-mates.

Defining a particular forward's contribution to a rolling maul is harder than assigning a home run to its owner for example.

Definitions can be subjective in rugby.

While no-one can dispute what a six consists of in cricket, whether a carry is 'dominant' or not can be argued. Opta surveys clubs with example clips, without always reaching consensus.

Cash, and rugby's relative lack of it, can also restrict what can be done.

Tracking data relies on ultra-wide camera angles, which are not available at every stadium.

Premier League budgets can offer the brightest talents more time and money to make breakthroughs and develop machine learning.

But the databases, and the thought about how to use them, are deep.

As are statistics' impact on scouting departments around the rugby world.