Sports buffs hit data-mining gold
Company that revolutionised analysis of cricket stats now contracts to government departments, agencies
Ten years ago, Jason Wells was watching the explosion of Twenty20 cricket, taking note of the futility of the sport’s traditional statistics to calculate the worth of the players.
“The way we measured players had to change,” says Wells, a former first-class cricketer. “Once Twenty20 came in it changed the way the game was played, so how you measured someone had to change. We realised it was all going to hinge on data.”
And from that epiphany we can fast forward a decade to the first floor of the old Dominion Post building, where he is co-owner and founder of Dot Loves Data, a firm that measures everything from the effects of weightrestricted rugby on player retention rates to the effect of pokie machines on urban deprivation.
So prolific has been their output that the company spent the last week at the University of the Sunshine Coast at MathSport, a biennial forum for Australian and New Zealand sports scientists to interact. There were 34 papers presented at the conference and Dot Loves Data had 14 of them.
Wells is co-founder and parent of Dot, along with fellow advertising old boy Matt West, IT specialist Mike Brough and mathematician Paul Bracewell, who comes from the famous cricket family (his uncles are former Black Caps John and Brendon, and his cousin is Doug).
They employ about 20 people across a small suite of businesses — Dot, EightyOne and Octave — from their central Wellington premises.
Dot is, West says, committed to being the best data company in the market, to growing and to Wellington.
“I reckon we’ll hit 25-30 employees pretty fast. I don’t know if we’ll stop there but what is more likely to happen is that we’ll carve off products and that might turn into a standalone 20-person operation . . .”
The shared passion, aside from their business and home town, is sport.
“In an ideal world we would do lots and lots of work in sport, but there’s no money in New Zealand sport,” says Wells. “But we’ve applied a lot of the smarts we’ve learned in sport into our data products.”
It is worth circling back to the birth of the T20 format to understand how data, if the stories behind it are told properly, can change long-held perceptions and shatter myths.
As the format gained a foothold it was assumed the best one-day cricketers would be the best T20 players. Batsmen were still being measured by their averages (runs per innings) and strike rates (runs per 100
A lot of people have used data to tell you what has happened but the real power of data is in prediction.
Jason Wells balls), and bowlers by the wickets they took and runs they conceded.
But the short, dynamic nature of the game meant that on any given day there was only really potential for two or three players to shine, so how could you measure the contributions of the rest?
Moneyball had laid a platform in baseball for managers and scouts to use metrics for previously unrecognised skills, and taking a deep dive into the data, Wells and co realised there was similarly untapped potential for T20. They realised the only statistic that really mattered was this: after you have batted, or bowled an over, does your team have a better percentage chance of winning? Some key points moved that needle — for batsmen it was the ability to hit boundaries off consecutive balls; for bowlers, dot balls.
“Everyone thinks a dot ball is boring but in T20 it is very, very important,” Wells says.
Inspired by the discovery that the most mundane facet of T20 cricket was its most valuable currency, it became the foundation stone of the nascent company’s name — Dot Loves Data.
Dot sold its T20 product to CricHQ (until it ran into cashflow trouble) and ex-Wellington Firebirds coach Jamie Siddons used it but it quickly became obvious that although sport was extremely data rich, it was, in New Zealand anyway, just as cash poor.
But the Dot team had gained confidence in knowing they could tell compelling stories through data.
“A big part of what we’re doing is trying to beautify data,” says West. “We have this picture of data scientists as these Poindexters in lab coats but it’s actually a really dynamic, exciting field that is just beginning to get tapped.”
Dot contracts mainly to government departments and agencies. It is working with ACC to predict injuries, types of injuries, and how much they will cost.
“That is fascinating,” Wells says. “To see if you get this injury, what is the next injury you’re most likely to suffer. Often it is the second injury that is the most costly.
“We’re using data to predict things, which is the big thing. A lot of people have used data to tell you what has happened but the real power of data is in prediction.”
This brings us to the product they’re most proud of — a deprivation index that updates monthly.
“When we update so often we can make predictions about what will happen to neighbourhoods in the next six months to a year,” Wells says.
“Most assaults occur between 11pm and 3am on Friday and Saturday nights. They’re predictable assaults because they occur within 50m of a bar and they’re also the most expensive because of the degree of severity.
“You can then start to see, if there’s a bar opening here, what we can do around that area to try to reduce the negative impacts.”
Dot is working alongside Hutt City Council to help implement strategies to minimise deprivation.
“It’s a pretty powerful tool when we can say to a council if you put a few pokie machines in, or take them out, what it will do for the long-term prospects of the neighbourhood.”
Interestingly, they’ve noticed one community input that has a massive positive effect on suburbs.
“Sports clubs,” Bracewell says. “If you want to improve your neighbourhood, put in a sports club.”