What sports analytics & insurance have in common

Brad Pitt sits in a movie studio talking to Jonah Hill about a made-up baseball team changing the game and winning 20 straight to surpass an all-time record. The movie – “Moneyball” – goes on to be nominated for six Academy Awards including Best Picture. It’s a fun and fascinating piece of fictional drama. Except that it isn’t. 

Before Pitt and Hill represented the Oakland A’s in the movie, “Moneyball” happened in real life and its narrative is considered to have largely followed the true story (also chronicled in the book “Moneyball” by Michael Lewis).

In 2002, the Oakland Athletics were an underdog to long-standing industry titans like the New York Yankees and San Francisco Giants. The odds were stacked against the franchise, but the A’s were still intent on trying to win. To do so, they had to get creative and consider defying convention.

What does this have to do with insurance? Every industry starts somewhere with data.

Power of data

The pivot to reach success was pretty simple for the A’s as they went about leveraging something at the disposal of every team decision-maker before them — data. Baseball has generated and consumed data for over 100 years. We can say with reasonable certainty that we know what happened between the Cleveland Spiders and Brooklyn Superbas of the National League in 1899 (the Spiders lost — they have the worst record in Major League history).

What General Manager Billy Beane and Assistant General Manager Paul Depodesta (“Peter Brand” in the movie) unlocked was the true power in the data as it came to the object of the game — scoring more runs than the opponent. The vast majority of what they believed in aligns with conventional wisdom (home runs are good), but some of it did not and it was in those margins, through the data, that Oakland won.

Twenty years later, we have reached commercial insurance’s Moneyball era.

Like baseball, insurance is also a tremendous generator and consumer of data, with some information dating back more than a century. It has legacy titans and legacy thinking that are ripe for disruption. It even has “losses.”

The time is now for the insurance data revolution

There are several reasons why it’s taken 20 years for data to make inroads into insurance. In sports, innovation is sexier and has a more immediate impact. Insurance is bigger and presumed to be less malleable than sports. Winning and losing are not as clearly defined. Technology advancements have lagged — especially with advanced data collection tools like telematics (relative to SportVue or Hit and Pitch f/x in baseball).

But now we are at the intersection of data, technology and innovation in commercial insurance.

Three ways commercial insurance is primed for its Moneyball era

The principle of ‘one number’

In the movie, Hill’s character states, “It’s about getting things down to one number. Using the stats the way we read them, we’ll find value in players that no one else can see.”

In insurance, that value is risk. Two major advancements are currently improving the industry’s ability to do this.

First, non-conventional data exists and is available (mostly in the public domain) to compare almost any attribute of a business to its history of risk. Rather than solely leveraging historical rating variables, a more complete picture of a business can be used to understand it.

Second, venture capital thinking has redefined what risk even means. In the SaaS (software as a service) markets where VC money has thrived for decades, lifetime value reigns supreme. This mentality translates to insurance by giving consideration to cost of acquisition, cost of service, likelihood to renew, customer experience, likelihood to cancel, collections probability and far more that may not always live in loss or combined expense ratios. This is the equivalent to focusing more on base percentage rather than batting average as a mechanism for run scoring in baseball (a core theme of Moneyball).

Understanding risk with both a more complete definition of what that means and picture of what that is for business can generate the same kind of marginal gains that Oakland uncovered. This results in an entire ecosystem – from policyholder, to distribution partner, to carrier, to investor that benefits.

Every attribute matters

For right around the last 20 years, online sportsbooks, brick-and-mortar casinos and even stadiums where events are occurring have offered in-game betting lines. This is the ability to make a bet on who will win during the game. In Europe and Asia, game betting makes up more than 70% of all money bet on games.

For about 15 of those 20 years, in game betting lines were incredibly exploitable because of how they were calculated.

To determine the likelihood of either team winning, a pre-defined algorithm would look at the history of all games to find past similar games and determine the relative chances for each squad. In football, a game in which the home team led by 7 with 14 minutes left in the fourth quarter with the ball around the 50-yard line would be compared to all other games like that ever. This ignores who is playing. Aaron Rodgers (back-to-back reigning MVP of the league) as QB for that home team as compared to his back-up (one career start) Jordan Love as the QB for the home team at that time should look totally different in those odds. The specific attributes of that game matter.

In today’s in game market, games are simulated from that point to the end of the contest based on all of the teams that are playing. This is less mathematically intensive and much more accurate.

A traditional actuarial approach in insurance can follow a similar path. All policies are grouped into homogeneous groups and reviewed across a large sample size of history to determine the likelihood and severity of loss.

Data – we know thousands of attributes about each business before it’s even underwritten – and technology can now converge to better understand what the drivers of loss are at a policy-by-policy level as opposed to rolled up to a homogenous group. Each factor (and combination of factors) that is known about a policy can be analyzed instantly for its risk. This allows for automation of underwriting (and even some actuarial math), which saves costs and significantly speeds up the process for all while also being more accurate in its understanding of risk.

Start with the box score

Moneyball focused on data that had been stored for hundreds of years. In fact, in 2018 the Supreme Court ruled that what happened at a sporting event was of public record and not data “owned” by anyone. Innovation could be done off the simplest forms of information.

In 2006, baseball implemented cameras throughout each of its stadiums to track the loft, distance, speed and even rotation of the ball and each of the players at all times (then called “f/x” and now run by the company TrackMan). Basketball did similar work with SportVue in its stadiums and football now has chips in shoulder pads. Sports has enjoyed telematics at the highest level for 16 years.

The following happened for the first time in 2019 (13 years later):

• The first game outcome prediction model based on telematics was proven to best similar models that solely leveraged what was in the boxscore (i.e. public domain).
• Football was able to reduce its margin for error in tracking players to less than +/- three feet (the data was previously often wrong by a full yard).
• Widespread adoption by fans, teams and players led to telematics data being used in broadcasts.

Telematics is the future — emphasis on future. We are just beginning the Moneyball era of commercial insurance. It will be important to collect and explore telematics data, but we are likely 10-15 years away from being able to leverage it in a way that adds value over the most mundane of data points.

Getting known, publicly available data to tell the appropriate risk story at the policy is the biggest step the commercial insurance industry can make in its marginal improvements.

There may not be entire television networks dedicated to covering insurance. Few people go around wearing Progressive t-shirts or AllState hats. Earnings calls from carriers do not draw tens of thousands of fans attending in person. Insurance does not excite people in the same ways as sports. Maybe that’s why similar innovation has taken longer, but the time for that innovation is now.

Paul Bessire has been working in predictive analytics for 17+ years. He has used data and technology to become one of the world’s foremost authorities at predicting sports and is now focused on doing the same with other industries. He currently works as the vice president of data for Coterie Insurance, an insurtech focused on making small business insurance fast and easy. Prior to joining Coterie, Paul was the VP of Advanced Analytics at AMEND Consulting where he helped middle-market clients – and some sports teams like the Cincinnati Reds and University of Cincinnati athletics – make more efficient decisions using data.

Opinions expressed here are the author’s own. 

Related:

• Digital trends are evolving in P&C insurance
• Can predictive analytics tame runaway verdicts?
• Fear keeps the digital insurance ecosystem from its full potential