How data science and Moneyball rebuilt Leicester Tigers

[InBetweenWasp]

Saw this article in The Telegraph last week.  Thought it was a really interesting insight into how Borthwick has sought to objectively look at how his squad is built and for recruitment.  Would seem to have given them a competitive edge initially.  Also has some references as to how/why they might be looking at Jimmy:

If and when the time comes to cast Steve Borthwick’s biopic, it might not be Brad Pitt that lands the lead role. But the remarkable Leicester Tigers turnaround, and the razor-sharp recruitment driving it, is more than a little evocative of Moneyball.

After inheriting a bloated squad in the summer of 2020, Borthwick has leant on data science to cut away dead wood, streamline the wage bill, survey the market and pick up gems amid shrinking finances. Although there are many strands to the club’s resurgence – not least the coaching of his charismatic backroom team and the emergence of young stars like Freddie Steward – such a rebuild has been critical.

It is in this respect that Borthwick is channelling Billy Beane, the former general manager of the Oakland Athletics baseball team played by Pitt in the 2011 film of Michael Lewis’ book. Guided by data, he is daring to think differently while coveting less-celebrated attributes.

Leicester are leaving the crowd behind, in more ways than one. They head into this weekend with an eight-point cushion at the top of the Premiership table and topped their Champions Cup pool. So far, the season has brought 16 wins from 17 matches in all competitions.

“Rugby has a lot of entrenched views that have become gospel,” says James Tozer, the co-founder of Oval Insights. “There’s that great quote in ‘Moneyball’, where Michael Lewis is paraphrasing Bill James: ‘If you challenge the conventional wisdom, you will find ways to do things much better than they are currently done.’”

'Too much money was being spent on players who weren’t performing'

Tozer worked with Borthwick at the Rugby Football Union. When the latter was approached by Leicester, he asked Oval Insights to reshape Tigers with their machine-learning models. His brief was to revitalise a giant of English rugby that had fallen asleep and crashed to the foot of the Premiership table.

A month ago, when asked about the current salary-cap investigation into Leicester’s historic dealings, Borthwick highlighted the “mismanagement on and off the field” that he and chief executive Andrea Pinchen had to fix. In Tozer’s words, there was simply “too much money being spent on players who didn’t get a lot of game-time, or who weren’t performing at a high enough level”.

Oval Insights’ models yielded expected points, or xP, which Tozer labels “universal currency that we can use to measure player contributions” in major competitions around the world. Rob Lowe heads up the data collection operation, plugging into the archive of Stuart Farmer. Tozer calls those two “the godfathers of modern rugby stats”. 

Ranked databases divided by position, with “exchange rate” adjustments for different situations, provide a launchpad from which to attack inefficiencies in recruitment, such as reliance on familiar networks or ignoring the full breadth of the market. Leicester’s management team comprises Borthwick, Pinchen, finance director Fintan Kennedy, head of recruitment Richard Wilks and general manager Leigh Jones. Tozer says the process is “very rigorous” and that analytics helps unearth “hidden gems”.

“In lots of cases, we’ve found players who are being released by clubs when they have strong xP numbers, often for doing quite subtle things, such as important defensive interventions or valuable kicks,” he adds.

“As a result, every member of Leicester’s current squad has at least one skill for which they are in the top 25 per cent of global top-division players in their position. They all have their own strengths, and the coaching team have taken them to new heights.”

It is this part that really chimes with the Moneyball method, which championed unsexy baseball attributes such as a batter’s tendency to reach first base. Tozer insists it would be unfair to disclose unique selling points of individuals, but Leicester’s recent loan signing of Olly Robinson from Cardiff Rugby reeked of logic. As statistician Russ Petty has highlighted, the 30-year-old back-rower pillaged 70 turnovers between the 2017-18 and 2020-21 Pro 14 seasons. That put him in the top four for each campaign.

There are further wrinkles. Tozer believes Tigers have been innovative in assembling versatile players capable of switching positions. As well as in-game benefits, this mitigates the possibility of specialists being starved of game-time.

“We have also done a lot of modelling on how players transfer from one league to another,” he continues. “If you look at Leicester’s recruitment in the last two years, they’ve drawn from a really wide range of backgrounds. Take Hosea Saumaki: he had never played outside of Japan, and has now scored match-winning tries in two big European games [both against Connacht].

“Another problem is failing to understand player ability at the start and end of a player’s career. Leicester have produced a brilliant crop of academy players recently, which has made a big difference. They have also been very good at identifying which guys in their 30s are still performing close to their peak, helped by the genius of [head of performance] Aled Walters.

“In general, younger players have less data, so are harder to evaluate. But, as a supplement to our algorithms, Oval has created a global panel of expert talent-spotters to help track players’ potential better. That idea is borrowed from ‘Astroball’, another baseball analytics book.”

Following Tozer’s point on veterans, it is difficult not to let the mind wander again. Tigers have been linked with Jimmy Gopperth, the 38-year-old Wasps centre who would provide a distributing midfield foil to incoming fly-half Handré Pollard. Undeniably, Borthwick has brought tactical pragmatism to Tigers. Tozer, though, is keen to stress that faith in xP, which marks players down for mistakes, does not outlaw creativity.

“I think people have believed that xP would discourage players from trying anything ambitious,” he says. “The reality is that the system rates lots of ‘flair’ players very highly… as long as they do more good things than bad overall!”

'Devoting a little budget to analytics can save millions on salary'

Earlier this week, upon his ascension to power at Worcester Warriors, Steve Diamond suggested that recruitment will be more important under a restricted salary cap. Leicester were ahead of the curve in this regard.

“Without a salary cap, using analytics intelligently can get you a long way, but there are still lots of other factors at play,” Tozer explains. “If you look at football, Liverpool have essentially been perfect in every major recruitment decision for the last five years, and still only won one Premier League title, because they can be significantly out-spent by other teams.

“Having a salary cap equalises things further, so intelligent spending becomes even more important. Devoting just a little bit of budget to analytics can potentially save millions of wasted pounds on salary.”

From next season, clubs around the world will be able to buy Oval Insights’ player analytics. That arm of the business will be overseen by Dr Mark Einhorn, who was hired from management consultancy firm McKinsey. The vision is for the machine-learning xP tool to become commonly used across the sport, part of an exciting future for the company. Already the official data provider to the United Rugby Championship, they will expand their inventory next season. Oval validate their xP ratings using Superbru, the predictions website, for URC fixtures.

“We’re beating 99.8 per cent of the 36,000 fans taking part, and outperforming all other public prediction algorithms,” Tozer says. “We also run the model at international level, and our predictions were much more accurate than the betting markets in the autumn. For example, we had Ireland to beat New Zealand by five, when the bookies were strongly the other way.”

Tozer hastily points out that neither he nor his Oval colleagues are permitted to punt on matches. With their help, though, Leicester have become far safer bets over a new Moneyball story.

“I think there are some parallels with the Oakland A’s,” Tozer acknowledges. “Andrea, Steve and the management team at Tigers have been forward-thinking enough to question the status quo of an inefficient recruitment market.

“They made a huge number of changes in other areas outside of data, too, and the coaching staff deserve enormous credit for the on-field performance. It’s really a much broader story of institutional reform, of which data is one part. Regardless of where the Tigers finish this season, the transformation that Andrea has overseen in two years is pretty extraordinary.”

Just as satisfying for Tozer, you sense, is that the ripples of rugby’s data revolution could reach far beyond the East Midlands.

“From a statistical perspective, I’m sure some rugby fans will continue to think that it’s a load of nonsense,” he finishes. “But for those who are open-minded, it would be nice if Leicester’s rebuild could be something of a turning point in how rugby thinks about data.

“Used in the right way, it can be pretty powerful stuff.”

[NellyWellyWaspy]

Thanks for posting. It still remains for the coach to make a teams of what he has, but it does make sense. The issue is that data is missing for players in teams where that data is not collected (like the Championship). There can be no doubt that this new lower cap has focused the mind on value for money. Which is why I still do not understand how we can afford to let Jimmy go. Just how many games has he won for us, yesterday included? We do not have an effective replacement for him.

[Rossm]

Thanks for posting. It still remains for the coach to make a teams of what he has, but it does make sense. The issue is that data is missing for players in teams where that data is not collected (like the Championship). There can be no doubt that this new lower cap has focused the mind on value for money. Which is why I still do not understand how we can afford to let Jimmy go. Just how many games has he won for us, yesterday included? We do not have an effective replacement for him.

Surely, until it is officially announced, then it still is only a rumour?

[Lwasp]

Hard not to be cynical about the timing of in depth reports like this. Tigers get accused of previous salary cap abuses so here comes an article about how the current management are focused on getting better players for less, "saving millions".

The thing people who quote Moneyball often forget is that whilst the As did so much better than previously, and expected, they still fell short of the big prize and then everyone else rapidly implemented similar programmes and negated the gains. Oval Insights will be keen to sell this system to as many people as possible quickly whilst Tigers are on top. Less impressive if they fade and don't win anything.

[InBetweenWasp]

Hard not to be cynical about the timing of in depth reports like this. Tigers get accused of previous salary cap abuses so here comes an article about how the current management are focused on getting better players for less, "saving millions".

The thing people who quote Moneyball often forget is that whilst the As did so much better than previously, and expected, they still fell short of the big prize and then everyone else rapidly implemented similar programmes and negated the gains. Oval Insights will be keen to sell this system to as many people as possible quickly whilst Tigers are on top. Less impressive if they fade and don't win anything.

Have to disagree, think it's easy not to be cynical and to easily separate what previous coaches and CEOs have done with salaries (which are being investigated) and how the current Coaches and CEO have chosen to handle recruitment and place values on players at a time when all teams are being tasked with providing more, for less.

That being said, I agree that we'll likely see Oval Insights being adopted by most, if not all Prem Teams soon enough in which case the advantage is diminished.  However, it's allowed them to get a head start by thinking outside the box and exploring an alternative way.  A bit like we did by quickly mopping up the cream of the Championship allowing us to get Hislop, Millar-Mills and Frost at what was likely incredibly good value.

Clearly, silverware isn't won by recruitment alone - you've got to get the talent to gel and coach them to actually win the games but interesting to read how another team - who also happen to have performed well above expectations - have used data to drive their recruitment process.

[backdoc]

I doubt very much if they would have been 16 wins out of 17 without Kevin Sinfield.

[Marlow Nick]

I'm intrigued how this works. My very simplistic and naiive assumption is that baseball is much simpler to analyse because it is to a large extent a series of individual performance "plays". Rugby analytics will need to factor in how players react to broken play and how teams can be bigger than the sum of their parts.
Interesting stuff!

[Rifleman Harris]

Notts County are owned by a Danish data company and they use data when recruiting players and the coaching staff! They have data on positional play (a forward being in the box when attacking) , contribution to the team (distance covered when the team is not in position etc), but that seems easier to measure than it would be in rugby. It seems to work reasonably well, but the coaching is still the major component, and personality.

[Heathen]

I doubt very much if they would have been 16 wins out of 17 without Kevin Sinfield.

Indeed. In the same vein, `i believe we are benefiiting from having Mitch on board. When we have 90+ % of our squad avaibale, I think we will be very hard to beat.

[Raggs]

Rugby is really hard to do with stats. There's so much on and off the ball to quantify. I'm sure a lot of it can be done, but it's also hugely situational, which just further muddies the waters.

I think stats is still a strong place to start though, since if a player isn't good enough at the important things to start with, the less tangibles don't matter anyway!

[Rossm]

Indeed. In the same vein, `i believe we are benefiiting from having Mich on board. When we have 90+ % of our squad avaibale, I think we will be very hard to beat.

Our defence has noticeably got much more organised and difficult to break down since Mitch has got hold of things with us.