Kelley L Cox-US PRESSWIRE
Andy traveled to Boston for the 2013 Sloan Sports Analytics Conference, and wants to get down some quick thoughts on the first half before moving onto day 2.
I always look forward to SSAC, mostly because it's the only time when I really meet people who work and think the way I do: super analytically about the sports. As a result, it's a whirlwind day, where I try to attend every event I can in order to learn as much as possible. After all, it only comes once a year. Here are my initial quick thoughts, and I'll do some longer, more detailed posts on these as necessary throughout the upcoming week.
- The conference is bigger than ever, with about 2,700 attendees this year. While it's easy to pine for the "good old days" of two years ago when the conference was half that size, restricting attendance too much probably isn't fair to all of those interested in this growing field. The size does, however, make it much more spread out and impersonal. It's about the size of my alma mater, Westminster, but without the 4 years of making friends first.
- Perhaps the biggest change that I saw was how much of a corporate feel this gave the conference. Sponsors "presented" a good portion of the panels, which tended to be much more stiff and formal than their unsponsored counterparts.
- Likewise, the companies with booths at the conference seemed to be larger ones (like HP, SAP, ESPN) than when I attended last. They also seemed to be selling largely business-to-business solutions: not much for the average consumer. It made that part of the conference a little bit less enjoyable.
- About 1,000 of the attendees were students, most of whom seemed immediately concerned about making a positive impression on important people. You would never have guessed the number of suits being worn (probably 50% of the crowd), and there were many business cards exchanged. I sort of hate that, and I imagine the important people do as well, but you can't blame all of the MBA/business major types for trying, I guess.
- They fed us lunch and had a cocktail party with delicious refreshments, both of which I am extremely grateful for.
- The Utah Jazz sent their 1st attendee ever to the conference. Adam Grow's job title is "Project Manager with the Utah Jazz" according to his LinkedIn page. He's been working with the Jazz for about 6 months.
- We (Benjamin Gaines of Salt City Hoops and I) had a chance to talk with Adam for about 10 minutes. He explained to us that he's on the business end of the Utah Jazz, not the basketball side.
- I had a brief chance to talk with Daryl Morey, GM of the Rockets, before he had to leave to make a call (perhaps to confirm signing Aaron Brooks?). I asked him about the Jazz' non-move of Jefferson and Millsap. He said, basically, that it would have made some sense to get some assets back for those expiring contracts, he doesn't know what the Jazz were offered.
- He does, however, feel that the Jazz are in very good hands with Dennis Lindsey. Remember, Lindsey worked for the Rockets organization before transferring over to be the Spurs assistant GM.
- In the first panel of the day, Morey talked about the difficulty of coaching a young team. In general, he said, they tend to miss pick and roll assignments more often than their veteran counterparts. The same is true for defensive rotations, young guys miss them more often than veterans.
- The good news, though, is that they're much more likely to get steals and offensive rebounds, plays which actually have much more importance than a missed rotation.
- Morey explained that coaches often struggle with young players because of this: they see the frequency of their mistakes, without realizing that it's being made up for in larger chunks.
- In theory, it makes a lot of sense to choose wisely when you go for 2, and when you go for 3. This is actually pretty intuitive: when you're losing, you should take more 3s than normal to get back into the game, but if you're winning, a steady diet of 2s should be enough to see you through.
- What Goldman and Rao found in their study, though, is that NBA teams do one half of this, but not the other. When a team is losing by a lot, they do indeed take more threes as they're theoretically supposed to.
- However, when they're winning by a lot, a team will actually take more threes than normal, which doesn't really make sense. Perhaps they feel more free to take those shots? Whatever the reason, they're hurting their team's likelihood of winning, albeit by marginal amounts.
- Perhaps the more interesting part of Goldman and Rao's study was the effect that being behind has on a team's performance. When losing, teams actually gain a .10 point per possession efficiency gain, even adjusted for the players on the floor. That's a really big deal in a league where .10 PPP separates the top and the average NBA teams.
- This is made up by two factors: a .07 PPP increase due to increased shooting efficiency, and a .03 PPP increase by continuing a possession on the offensive glass.
- Why is this happening? Well, the authors don't explain it, but I would guess it's due to one of two causes. One, perhaps the inefficiencies above account for the gap in shooting. 3s are usually higher efficiency, and remember, teams trailing do indeed take more 3s, as they should. Or two, what I call the "Dick Bavetta NBA Refereeing Fairness doctrine". Basically, refs are more kind to the trailing team, leading to more FTs, and fewer loose-ball foul calls on the losing team. There is some, but pretty limited, evidence for this conclusion.
- Since trailing teams come back so often, it's actually true that the 4th quarter is more important than those which come before it. Likewise, the Jazz deficits in the 1st probably matter less than they otherwise would.
- I want to do a longer post on this one, but this was a very interesting post using the SportVu XY camera data. Basically the study's authors, Jenna Wiens, Guha Balakrishnan, Joel Brooks, and John Guttag, looked at whether or not teams should crash the offensive glass after taking a jump shot, or if they should instead get back on defense.
- The decision is not an easy one: teams who send 3 players to crash the glass get the offensive rebound 40% of the time and only 5% when they retreat 4 players back on defense.
- On the other hand, sending at least 2 players to retreat means that you're 4 times less likely to give up a transition bucket on the other end.
- In the end, though, the study found that crashing the glass was more effective: sending just 1 additional rebounder to crash gave the team a 0.17 point advantage per opportunity, totaling to just over 4 points per game. That's a big deal.
- I suspect, somehow, that this study doesn't consider everything, mostly because that difference just seems too important. More on that in the full post.
- I didn't actually interact with Rovell today, or even hear him speak, but I still don't like him.
- I did see him, though, I'm sure he spent hours on his hair.
- Kirk Goldsberry (the CourtVision guy who writes for Grantland) and Eric Weiss continued their use of their SportVu data to study effective post defenders. Essentially, they looked at what happens when a post player is in the restricted area: how often does a paint field goal actually get shot when a particular big man is down low, and if it is taken, how often does it go in?
- Dwight Howard and Larry Sanders turn out to be excellent paint defenders. They do this in different ways though; Howard dissuades the opposition from taking shots in the paint at all, while Sanders simply lowers the percentages of the shots that come from in the paint.
- On the other hand, David Lee and Anderson Varejao look pretty bad by this metric: they allow both a higher FG% and more shots in the paint than usual.
- Al Jefferson, the only Jazz player with enough of a sample size, was in the middle of the group. He ranked 31st out of 52 players in terms of paint allowed FG%.
- Almost universally considered the next big evolution in sports stats, XY data from STATS, SportVision, and other companies use cameras and video recognition to track each player's, referee's and ball's location 30 times per second.
- In basketball, the system is STATS, Inc.'s SportVU, which installs cameras up in a stadium's rafters to track the objects on the court and their movements. STATS sells the systems to NBA teams for $70,000 to $100,000. 15 NBA teams currently have the cameras, the Jazz are not one of them.
- In the panel discussing this data, a VP from STATS, Brian Kopp, explained that the reason this data wasn't universally available (to both public and team analysts) was because (not a direct quote) "We can't just break into the stadiums and install the cameras, we need to work with the teams in order to be able to do so."
- Let's be honest, that's absurd corporate PR. Teams would be happy for you to enter their stadiums and install the cameras, unfortunately, you charge tens of thousands of dollars to do that. It's not universal in order to protect your income stream in what is an incredibly limited market of 30 NBA teams.
- That's a bummer for analysts, for reasons best explained in this Slate article.