For the second straight year, I was able to venture up to Boston for a weekend to attend the biggest gathering of famous sports minds, ridiculously smart people, and well…nerds that I’ve ever been to. It was in a new location with a record crowd and is really setting the bar high for the advancement of analytics in sports. As one of over 1,000 students in attendance I went to panels covering a myriad of sports, ideas, and tools and their relationships to data and analysis.
Throughout the panels there were several reoccurring themes I learned. Here’s my take on a few of them:
Quality of Analytics vs. Presentation of Analytics
A major topic was the disconnect at times between the public and analytics. Not only is there a misconception that “analytics” have to be crazy formulas, numbers, and math (they don’t!), but there is also a certain stigma to statistical analysis from some coaches, managers, players, and fans. The Mike Trout/Miguel Cabrera argument was the perfect example. Each side thought they were so right that it created a sort of stalemate that hurt both sides’ arguments – neither could bear listening to the other. This all leads to the fact that having an analytic tool that makes sense and is presentable is just as important as the quality of the analytic itself.
49ers COO Paraag Marathe said, “A 5/10 piece of analysis is much better than a 10/10 piece of analysis if it’s conveyed in a succinct way.” With more and more analytics in the public eye this should be getting easier, but still has a ways to go. It’s one thing with a stat like on base percentage, where it is exactly what it’s name says and very simple for any fan to understand. It’s when you get into a stat like WAR with tons of variables and no clear definition to the average fan that trouble arises.
There was even a whole panel titled “Beyond Crunching Numbers: How to Have Influence!” While I didn’t attend that one, its sentiment was echoed by many other panels and speakers. This was definitely a key lesson to learn as we continue to try and produce advanced content at RotoAnalysis – you shouldn’t keep it simple and baby your reader, but you should definitely work on your presentation so that it is understandable and direct.
Process vs. Outcome
Another major misconception in the public is that regarding variance and “luck.” The classic example is that coin flips won’t always just be 50% heads and tails, and in sports that is perfectly applicable to a shot, game, or even season in some cases. The Baltimore Orioles were 29-9 in 1-run games last season and 16-2 in extra innings; both are absolutely ridiculous marks. They were clear signs of regression and that their team that maybe should have gone .500 ended up .574 on the year. Vegas has responded by expecting them to put up a .472 winning percentage this year.
Nate Silver explained, “The short run is luck dominated, and the long run is skill dominated.” This kind of analysis is the best way to show that the process, not the outcome, is the best predictor of future success and the best analyzer of how true a certain happening is. Farhan Zaidi, the Director of Baseball Operations for the Oakland A’s, said, “Make the analysis of skill instead of outcomes in all measures.” By examining the process in which you make your decisions, you’ll be able to tell whether or not you took the right steps.
Celtics Assistant GM Mike Zarren took this even farther, agreeing, “you can clearly do everything right and not win, but you can’t do everything wrong and win.” This shows exactly the process you need to understand things in. Variance is a huge of sports and always will be – that’s what makes it fun and us obsessive about it.
The Media Matters
I am paraphrasing, but one of the key points Mavericks owner Mark Cuban made about ownership was that the community owns the team, not just the owner by his or her self. ESPN President John Skipper said, “If I was going to be handed a sports franchise, the first I would spend my time thinking about is the fan experience.” With panels about “fanalaytics” and ticketing the relationship between teams and the public is very important, and the third party that is the key to their harmony is the media. The battle between analytics and the history of how sports “work” has been targeted the media for years, and many team representatives touched on this.
Voros McCracken discussed one year when he was working for the Red Sox where they wanted to be more creative with their bullpen. Without a true closer, they would be playing the matchups more than anything and use their best guys in the highest impact situations (even if not in the 9th inning or a save situation). Their bullpen wasn’t bad, but the media didn’t understand their strategy and trashed the team for lacking a closer. They ended up making a few meaningless trade deadline moves to switch around their bullpen in an effort to simply quell public distaste.
This wasn’t the only example of how media impact can hinder creative strategies. The “Monday Morning QB” panel had two former NFL head coaches, Herm Edwards & Jack Del Rio, contrasted with Falcons GM Thomas Dimitroff and the founder of Advanced NFL Stats, Brian Burke. Polls from the audience were incorporated to a program that went through old decisions of coaches and broke them down from a statistical, social, and football perspective. It was clear that throughout games, how the public will react is considered.
Sports Are the Perfect Case Study
Pure data in sports is easier to find than in many other businesses like finance. There were a ton of interesting ideas thrown around throughout the weekend regarding applied mathematics, statistics, and sports merging together. Among these was Joe Peta presenting a new foray into asset management: a sports gambling based hedge fund.
Many panelists, and Nate Silver in particular, said gambling and fantasy sports are two ways to experiment with these forms of mathematics and statistics that can quickly be converted to other industries and methods. Silver was actually a part of the poker boom in addition to his baseball work before becoming the political analyst named one of Time’s Top 100 Influential people and said, “Poker players are the best I’ve seen at evaluating situations correctly.” Even though this conference may be pigeon-holed into the category of sports, there are lessons for nearly any profession to be learned, and representatives form many major companies like SAP, Citi, Credit Suisse, and other businesses were also in attendance.
We’ve Got a Ways to Go
The final theme of the conference was that despite all of our progress, we are still only scratching the surface. Baseball is a sport with a lot of information out there and probably the most advanced, but is not reaching its point of diminishing returns just yet, the “Baseball Analytics” panel agreed.
Basketball next up in terms of the analytics revolution; some of the XY-data from SportVu and new data around the sport is just fascinating. Kevin Pelton, a former RotoAnalysis podcast guest, had a nice piece on 10 questions to be solved in the NBA. Many of these were touched throughout the conference for basketball and other sports, and some of the more interesting and more important possibilities include:
- Injury Analytics
- Scouting Analytics
- Team Chemistry Analytics
- Coaching Analytics
- Spatial Analytics
As analytics move forward, these will hopefully be tackled by some of the smart minds that were in attendance this past weekend. I had an amazing time and encourage any sports fan, regardless of passion for analytics, to attend.
Matt Cott is a co-founder of RotoAnalysis and current student at the University of Virginia. Follow his work all season long on RotoAnalysis.com & CBSPhilly.com and direct any questions or comments below or to his Twitter @KidCotti21!
 A really smart guy who I had the chance to ask a couple questions outside of a panel. He should be around the NFL for a long time, and is a definite reason for the Niners’ strong organizational makeup right now.
 If you haven’t heard of him, you’ve heard of his stats. The dude essentially invented all defensive independent pitching statistics (DIPS).
 Another highlight of the panel was Tony Reali moderating. He was sharp and funny and produced one of the more thought-provoking ideas I heard all day. Say you’re down 3 on your opponent’s 40 yard line. Not sure your kicker can kick a 57 yarder? Ice him. Yup, ice your own kicker just to be able to see if he can reach on his first kick. If he can, kick it, if he can’t, go for the hail mary. Makes too much sense and stemmed off of a whole discussion out of how stupid icing is. Maybe it could be useful after all.
 The winner of “Evolution of Sport” addresses, by Adam Guttridge, was an automated prospect model that was very interesting and a potential baseline for future scouting of baseball prospects. Very nice complement and far more detailed than my top 125 prospects.
 Kirk Goldsberry of Grantland had a great research paper on “The Dwight Effect,” where he studied interior defense when players were within a certain distance of the shooter. The next focus of NBA analytics is definitely evaluating these sorts of data.