The advent of in-depth analyses of football statistics, led by the website Pro Football Focus (PFF) and its football fan founder Neil Hornsby in 2004, changed the way that teams evaluated their players. Instead of just assessing touchdowns, yards and catches, PFF broke down the complex system of individual behaviors and interactions – from the first step a player takes when the ball is snapped to whether they can effectively executive a countermove – and provided teams with a deeper understanding of each player’s performance.
With 3,000 evaluations per game and over 750,000 evaluations per season, PFF aggregated metrics that were more useful than those traditionally used, according to Grady Irey, Arity’s vice president of data science, in his recent blog post, “Looking Outside In: From Luton Town, England to the New England Patriots.”
For the VP, there’s a clear connection between the NFL’s use of analytics and better auto insurance rating systems.
“In the world of predictive analytics, high frequency events tend to provide a clearer signal,” he told Insurance Business. “To use an insurance analogy, things like catastrophes and hurricanes happen with a very lower frequency unless you look at a very long period of time, so they’re actually pretty hard to predict.”
Car accidents, too, tend to be low frequency events, which makes it difficult to determine the risk associated with each individual driver.
“The average person only has a claim every five to 10 years, depending on how risky that person is, so that’s by definition a very low frequency event,” said Irey. “What insurance companies have done historically is to look at things they have found to be predictive that happen a little bit more frequently,” such as tickets and moving violations.
Even then, many people don’t get tickets for years, so insurers have buffered the limited information with data comparable to what the NFL initially collected, like measurements of height, weight, speed, and strength.
“In the insurance world, that’s like how many years have you had a driver’s license and how old are you, and in states that allow it, what’s your gender and your marital status, and what does your credit report have on it,” explained Irey, adding that the information is useful to a degree, but it doesn’t truly model who somebody is and how they drive, which is where telematics comes in.
“When you use telematics, you can actually see that fine, granular-level of detail and illustrating things that happen, not only every day but multiple times a day, and that is far more insightful than something like a moving violation or an accident that might happen every five or 10 years.”
The applicability of telematics is painting a clearer picture of drivers and their risk, and even though larger insurance companies have led the way in adopting the technology, Irey predicts that everyone will be jumping behind the driver’s seat in the future.
“The higher market share insurance companies have gotten into telematics much sooner, even before they really knew that there was going to be a near-term payoff in revenue or profits,” he said. “Over time we’ve seen more and more carriers getting involved, and some of that has to do with telematics service providers and analytics service providers making the insight and the data origination a bit more approachable for companies of smaller scale.”