Thinking Outside the Strike Zone: Actuarial Science & Baseball Analytics

Twenty-five years ago, just five years after receiving his Fellowship in the Society of Actuaries, John Dewan, FSA, left a highly successful career as an insurance actuary to pursue a life-long dream, the development of the most timely and comprehensive computer database in sports. This became STATS, Inc., a sports statistics company that was eventually sold to Rupert Murdoch and FOX News for 45 million dollars. John is the author of four volumes of The Fielding Bible and the owner of Baseball Info Solutions, which provides analytical services to 22 of Major League Baseball’s 30 teams.

We talked to John about his background as an actuary and how he became so interested in baseball stats. If you’re interested in learning more, John hosted a webinar on Predictive Analytics and Baseball on September 27th, 2016 (click here to purchase the recording). His last webinar, an overview of Actuarial Science and Baseball Statistics, is also available on ACTEXeLearning, click here to view more. The webinar will cover how improving data collection changed the sport of baseball, specific examples of the applications of data mining and actuarial science to baseball, areas of overlap between actuarial science and sports analytics, and how you can get involved in the growing industries of sabermetrics and sports analytics. This is the first webinar in a 2-part series. The second webinar in the series will go more in-depth into how predictive analytics can be applied to baseball.

 

AZ: How did you become interested in baseball statistics and research?

dewan
John Dewan FSA

JD: I grew up on the south side of Chicago. My dad always had the White Sox on the radio no matter where we were and I became a baseball fan and a White Sox fan.

My interest in baseball statistics started when I was eight when I learned to keep score at the very first major league game I ever attended.  It was, of course, a bat day double header.  They described a scoring system in the program/scorecard, a system that I still use to this day when I go to a game.

Then it was Strat-O-Matic baseball on the steps of my front porch, a la Spike Lee.  I was “Bowie” Dewan, named after the commissioner of baseball, Bowie Kuhn.  I organized the neighborhood league, and when we weren’t at the playground playing baseball, we were on my front porch “playing” baseball.  I learned so much from that board game; I still play the computer version of the game to this day.

It was from Strat-O-Matic that I learned the value of a walk.  I was in a league once where I drafted players based on batting average.  I had learned from all the baseball announcers that the players with the best batting averages were the best players.  I got trounced in the league, while the guys who had players with a lot of walks and home runs won everything.  I remember the day when I came home from a losing day playing Strat-O-Matic and invented my own formula to rate players (very similar to today’s OPS) to incorporate the value of walks and power.  The formula that I still use to rate Strat players is: (Batting Average + 3 x On-Base Percentage + 3 x Slugging Percentage)/7.

AZ: Where did you go to school and what did you study?

JD: Loyola University of Chicago. I met my wife at Loyola. We both have degrees in Math – we met in math class! I also co-majored in Computer Science.

AZ: What led you to take the leap from working as an actuary with a part-time role in baseball research to becoming the CEO of STATS?   What did you see in STATS at the time?

JD: I was working as an actuary in the early 1980s and a co-worker gave me the Bill James Abstract to read.  I read it cover to cover in one sitting and was astounded.  Bill James was doing baseball analytics with baseball statistics in the same way that I was doing insurance analytics with insurance statistics.  That was the turning point of my career.  I loved my work as an actuary but I knew that I could have much more fun working with baseball numbers than I could working with insurance numbers.

A couple of years later Bill wrote about starting Project Scoresheet. I remember sitting at the kitchen table reading about Project Scoresheet.  I was mesmerized.  Now he was talking about what I dreamed about doing.  We had computer databases at the insurance company and I’d dreamt about having a baseball computer database.  Bill was talking about a grass roots project to collect baseball scoresheets.  I wanted to be part of this; I wanted to computerize the collection process.

I stared into space for a minute or two and then got up and dialed directory assistance for Bill James in Lawrence, Kansas.  To my surprise, I got his phone number.  I spoke to his assistant, Jim Baker, who got me hooked into the project.  Within a month I was writing the code for the collection software, and by the next season I was the Executive Director of Project Scoresheet.

While directing Project Scoresheet, I hooked with Dick Cramer, who wanted to restart the failed start-up of the first version of STATS.  We co-founded the second version of STATS, Inc.  Dick had developed collection software that collected data at a deeper level than the software I developed.  Namely, pitch-by-pitch collection.  I was very intrigued about developing a deeper level of baseball data at STATS.

 

baseballAZ: Why did you decide to return to baseball stats with Baseball Info Solutions?

JD: After we sold STATS to Rupert Murdoch and company in 2000, I remained as CEO but I was primarily interested in research.  There was little interest in doing the kind of research I was interested in doing at STATS at that time, so I left after a year.  We then started Baseball Info Solutions after my one-year non-compete expired.   We would go even deeper now, not just collecting pitch information like “swinging strike”, “ball”, “called strike”, but also collecting pitch type, location and velocity.  It was a huge breakthrough in the baseball analytics industry.

AZ: Your research now focuses on baseball defense. What in particular interests you about baseball defense?  What is the most challenging aspect of measuring it accurately?

JD: My interest in defense is based in two main things: 1) Strat-O-Matic rated players defensively and I wanted to see if there was a way to add some objectivity, and 2) because there was no good way of measuring defense in a good way, I wanted to develop better methods.  Fielding Percentage as a defensive stat was even worse than Batting Average as an offensive stat.

One of the first things I did when we had Project Scoresheet data was to present defensive analytics at the SABR convention in 1986 when I showed how the new data could help evaluate players defensively.  I focused on how pitcher handedness and groundballflyball tendencies could bias range factors.  I developed a new version of range factors that mitigated these biases.

Than at STATS, we had better data.  We had zones that we recorded for where each batted ball was hit.  We developed Zone Ratings, and then Ultimate Zone Ratings.  We published these analytics in the annual Baseball Scoreboard.

Then at BIS, our hit location data was even better than at STATS and we developed the Plus/Minus System, which then led to Defensive Runs Saved.

The biggest challenge with measuring defense is that there are so many aspects of defense.  For example, for catchers alone we measure how well they throw out base runners, how well they prevent stolen base attempts, how well they block pitches from becoming wild pitches, how well they block the plate when a runner is trying to score, how well they handle the pitching staff, how well they frame pitches to get or lose extra called strikes.  One of the keys is tracking more and more data elements that enable you to look at all these things. We are constantly adding more elements to track at BIS. We are getting better and better all the time, but there will always be more things we can do to measure things analytically.

AZ: Thanks for sharing John!

John Dewan is leading a webinar on Actuarial Science and Baseball Analytics through ACTEXeLearning on April 5th. This is the first webinar in a 2-part series. The second webinar in the series will go more in-depth into the field of predictive analytics applied to baseball.

Purchase the recording of John Dewan’s recent webinar on Predictive Analytics and Baseball.

Watch John Dewan’s first ACTEX Learning webinar at www.actexmadriver.com