Kiwi data genius crunches the rugby numbers
Dr Paul Bracewell is described as ‘‘a genius’’ by Ken Quarrie, New Zealand Rugby’s chief scientist. He also has sport in his blood. His father, Doug, played rugby under Sir Brian Lochore at Wairarapa Bush. John Bracewell is one of his uncles, one of the four Bracewell brothers who either played first-class cricket or rugby, with two of them playing test cricket for New Zealand.
And Dr Bracewell has also produced a groundbreaking model for rating more than 3000 professional rugby players across the planet.
It is science meets passion meets family history and it is an absorbing piece of work that now sits on The Rugby Pass website for anyone to use.
‘‘How far back do you want to go?’’ Dr Bracewell asks when quizzed about how it all started.
‘‘I’ve been in data for the last 20-odd years so for me personally it started when I went up to Auckland to chase a dream to try and play first-class cricket, which never eventuated.
‘‘But while I was at university I took a couple of stats papers and thought, ‘Wow, this is awesome, this is really interesting what you can do with numbers’.’’
As his sporting dreams faded it was the eureka moment that told him his contribution to sport could lie in another area: the growing field of analytics.
‘‘I did my Masters looking at cricket stats and my dad said, ‘Why don’t you have a look at rugby stats?’ and so I did.
‘‘My PhD was looking at creating rugby player ratings and this is going back to the late 90s-early 2000s
‘‘And that’s an interest that has never ever left.’’
Like many complex projects Dr Bracewell’s work has a simple question at its core: can you objectively measure the performance of individuals in a team sport?
The answer to that is yes; by quantifying the moments that really help teams win.
‘‘It’s the heart of it,’’ says Dr Bracewell, who is a partner at Wellington company Dot Loves Data. ‘‘These are team sports. You are doing something for the greater good within the team context so what do you need to do in any point in time to help your team’s chances of winning?
‘‘So, that takes into consideration the match context.
‘‘So, if you drop the ball at nilall three minutes into the game that doesn’t mean the same thing if you drop the ball with the line open with one minute to play and you are down by three points.
‘‘The challenge you’ve got is players playing in different positions, different roles, different match structures – how do you go through and understand what they were trying to do and how they contributed to that sequence of play? So being able to isolate those key moments, you can start to see how you contributed to how well your team did in that phase of play.’’
Values are allocated to the many actions a player takes during a game and after all this data is collected and processed, each player is given a final score. As a result it’s a model that rewards players who, in simple terms, ‘‘win the big moments’’.
That’s why Richie Mo’unga, Kieran Read, Jack Goodhue, Ardie Savea and Codie Taylor are rated as New Zealand’s most influential players in Dr Bracewell’s model.
It gets technical, of course. Dr Bracewell explains that the ratings are based on data collected over a two or three-year period and are forward-looking, helping to predict how player X is likely to perform the following week.
Like most data scientists Dr Bracewell talks with missionarylike zeal. It’s not that he pretends to have all the answers but he clearly cares deeply about trying to make sense of the games we watch every weekend.
And what of Quarrie, who has been recognised by coaching legend Wayne Smith as one of the reasons the All Blacks have kept one step ahead?
Would NZ Rugby have a similar model to Dr Bracewell’s stashed somewhere in a secret bunker in its Wellington HQ?
Dr Bracewell laughs and if he knows he isn’t telling.
‘‘They’re doing some very, very smart stuff at NZR and Ken is a very, very smart man. And we’ll probably just leave it at that.
‘‘He’s a real asset for New Zealand sport.’’