I spent the past weekend at the 2014 Sloan Sports Analytics Conference in Boston. It was my fifth straight year attending and, as always, there are many interesting matters discussed as it pertains to the use of statistics in sports.
While analytics if the focus of the conference, there are always some discussions that are more wide-ranging in nature, so I'll try to cover as much as I can, with this write-up providing a more general overview of the conference before I post the second part, which will focus more specifically on the hockey content.
NBA COMMISSIONER ADAM SILVER
NBA Commissioner Adam Silver was impressive in a one-on-one interview with Malcolm Gladwell. Gladwell grilled Silver and, for the most part, he was forthcoming with his responses.
Now, when Gladwell hammered away at the greed of some NBA owners, foremost among his concerns was the Dolans in New York, getting tax breaks as they aren't required to pay property taxes on Madison Squate Garden, one of the most valuable pieces of land in Manhattan.
For as much grief as Gladwell takes in some quarters for the pop science in his books, he was a great choice to interview Silver because he wasn't afraid to ask the commish difficult questions, yet did so, with follow-up queries, without the interview getting uncomfortable or contentious.
Some of that goes to Silver, who was forthcoming in his responses or, when defending the tax positions of billionaire owners, was clearly speaking from the commissioner playbook. There are times it's transparently difficult to defend the indefensible.
Silver talked about the removing disincentive for teams to win and this was a big topic at the conference. It speaks to the situation in which the NBA finds itself this year, with a handful of teams competing for the title and perhaps twice as many aiming for a top lottery pick.
Houston Rockets GM, and Conference Co-Chair, Daryl Morey plainly stated that the NBA has to find a way to eliminate tanking incentive.
New Sacramento Kings owner Vivek Ranadive joked with Boston Celtics owner Wyc Grousbeck, on a panel about Ownership Perspective, that Silver said the Celtics were tanking, to which Grousbeck quickly replied, "And what you are you doing?" The Kings are currently 23rd in the NBA standings, the Celtics 27th.
Former Toronto Raptors GM Bryan Colangelo even caused a stir when he admitted to tanking. "I tried to tank a couple of years ago," he said on the In-Game Innovations panel. "It didn't work." Colangelo made it clear that he never told player or coaches not to do their best to win but, as anyone seeing how teams are constructed would understand, management can put together a roster that is less likely to win games.
Former coach Stan Van Gundy is not a fan of the tanking. He ripped the Philadelphia 76ers, with Sixers GM Sam Hinkie in the audience, saying, "What the 76ers are doing is embarrassing. Can't put out that lineup and think they can win."
Boston Celtics Assistant GM Mike Zarren, a panelist every year at the conference, has proposed a wheel format that would replace the current draft and it's a fascinating theory, definitely one that would remove the incentive for tanking, but there are practical issues at play too. Silver loved the idea initially, but acknowledged that there will be some hurdles to overcome if that kind of solution is going to happen in reality.
Silver also mentioned how NBA players have approached the league about a mid-season break for all, noting how All-Stars don't get a break and the grind of the season can be taxing on the league's best players, logging more than 30 minutes a night, night after night. This was particularly interesting, given all the recent complaints in NHL circles about the Olympic break.
The most shocking admission from Silver was that he acknowledged how gambling and point spreads increase interest in games that might otherwise be mismatches. This is no secret to the world at large, but the nature of all league commissioners has been to dismiss out of hand the benefits that the league might receive from gambling.
When discussing how much teams ought to be spending on analytics, Colangelo put the range between $250,000 and $500,000 per year. When you consider the size of contracts in professional sports, it's not unreasonable to think that an analytics department could lead to decisions that would save at least that much.
One of the great developments in terms of NBA statistics is the emergence of SportVu data. Among the countless things tracked by the SportVu cameras, they can track how far a player has run -- Trail Blazers SF Nicolas Batum is up to nearly 160 miles for the season -- and Grantland's Zach Lowe, moderating the Basketball Analytics panel, wonders if that kind of data will really prove to have value in terms of injury prevention. The detail of this tracking data is so new that it's difficult to know, but it's easy enough to see how, someday, knowing the specifics as it relates to a player's workload could result is more efficient usage in order to reduce the chances of injury.
To counter, Van Gundy vehemently disagreed with playing time limits because he felt it was a lot of guesswork and cited the Washington Nationals' decision to shut down Stephen Strasburg in 2012 as a random determination. "How did they come up with 160 innings," asked Van Gundy. "Why not 170?" Van Gundy insisted that professional sports come with a risk of injury and that not playing obviously reduces that risk, but only because no one will get hurt when they aren't actually participating.
As the old-school NBA guy, Van Gundy also insisted that there is no substitute for watching film, and that the only numbers he trusts are ones that his own people keep. The trouble with this approach, of course, is that if the only statistics that matter are kept by internal sources, then biases are more apt to creep in than if it is a league-wide measurement.
No one pointed this out to Van Gundy, but it could also be said that the value of statistics isn't so much in learning more about your own team. It can certainly help, but helping to provide value and context for players throughout the league is where statistics can provide much more information. As ESPN's Director of Production Analytics Dean Oliver noted on another panel (re: college football playoffs), "A human can see a game better than the numbers can, but the numbers can see every game."
While I use some of Van Gundy's comments to echo a point made by Steve Kerr, the former Suns GM, who noted that getting buy-in from some coaches remains an issue, it's not as if Van Gundy is a total dinosaur and it's enitrely possible that a coach views the game differently than those tasked with building long-term. As Van Gundy stated, "When you're coaching, you're coaching to your personnel, not to an analytics philosophy," and there's some measure of truth in that, because there's no point in having an analytics stategy that doesn't fit your personnel.
As part of the In-Game Innovations Panel, George Karl was quite progressive in his thinking, but acknowledged that he had a track record that allowed him more flexibility, contrasting with younger coaches who can get crucified (by media, fans) for trying to be innovative.
Misaligned goals is a common problem for professional sports teams and it's part of the reason that there was a panel at Sloan this year entitled Inside the War Room: Building Alignment from Front Office to Field. It featured Atlanta Falcons GM Thomas Dimitroff and Heach Coach Mike Smith and it's an issue that is pervasive when it comes to applying analytics to professional sports.
It's one thing to have Mark Cuban as your owner and the organization falls in line because the owner takes a clear position and it's another to have a general manager trying to keep his job and a head coach trying to keep his job and they aren't necessarily pulling in the same direction to achieve their respective goals.
Furthermore, as Boston Celtics coach Brad Stevens noted, "We live in a day and age with so much more information," he said. "The challenge is communicating that information to the team." Stevens admitted that some players, like Rajon Rondo, would devour the analytics because that's the way he's wired ("He could be here," Stevens joked.), but that other players aren't wired that way, so it falls on the coaches to try to get their point across in a way that will appeal to players on their terms.
It's not easy to get players to buy into stats. On my flight home from Boston, I happened to read a feature on Tom Brady during which Brady made a couple of seemingly contradictory statements, one of which was, "I would say I perform at a much higher level as a player now than I did when we won Super Bowls." The other was: "There's one stat that matters, and that's wins."
When you're a player, wins are what matter; that's the objective. What analytics will tell us, though, is that Brady is precisely right that he is playing better now than he did when New England was winning Super Bowls, the last of which came after the 2004 season. That his team hasn't won the championship does not mean that Brady's individual performance has declined, a position that is taken altogether too often when evaluating the performance of individual players. A lot that is beyond the control of one individual goes into whether a team wins or loses a game, let alone a championship.
As the Godfather of Analytics, Bill James offered a measured response for coaches, particularly as it applied to in-game innovation. James said that coaches are control freaks, but chaos can be useful, so the best coaches learn to appreciate the chaos. This was followed by Morey telling a story about when Jeff Van Gundy was the coach in Houston and Morey had printouts for all the plays that the Rockets would run, with measures for efficiency and Morey didn't recognize the code for the most effective play, so he asked Van Gundy.
"That's Random," said the coach, "when the (called) play breaks down." Sometimes all the analysis and planning goes out the window and you just have to let the players play.
Stepping aside from the NBA for a bit, it comes as no surprise that baseball would be well-represented at a sports analytics conference. After all, baseball brought us the aformentioned Mrl James and Moneyball -- the kind of objective analysis that is at the core of modern analytics.
The baseball panel, which was likely the best panel of the conference -- at least the best I saw, using my Panel Analytics -- was moderated by MLB Network's Brian Kenny and that set the right tone. Kenny favours an analytical approach and the panel included Vince Gennaro, Bill Squadron, Rob Neyer, Nate Silver (of 538 fame) and Houston Astros GM Jeff Luhnow.
To no one's surprise, a panel full of analytics-oriented people had an interesting conversation about advances in analytics.
Even at a point when Kenny ventured towards the idea about whether there would be a way to actually measure chemistry, he did so with some obvious skepticism and I was in full agreement with Neyer's point that there are plenty of teams with great chemistry that don't win and teams with horrible chemistry -- citing the Oakland A's and New York Yankees of the 1970s -- that won a lot. My take is that chemistry is the kind of thing that gets awarded to winners after the fact. Sure, the Red Sox won the World Series because they had great clubhouse chemistry, but would that chemistry have been any different if they didn't win?
Luhnow later said, "There is no correlation between being a nice guy or a good person and being a good baseball player." The character issue is a tricky one when it comes to sports. No one wants to work with someone that is a pain in the butt, but if that person can help you win, sometimes the decision isn't so easy.
Neyer also noted how economics haven't quite caught up with analytics. "Does anyone get paid for pitch-framing," he asked, addressing one of the hottest evolutions in defensive data -- the catcher who saves his team by drawing more called strikes. Neyer asked if anyone was paying for Jose Molina saving a team 20 runs with pitch framing and while the flat answer is no, there are some players that have major league jobs because they are adept at pitch framing. Of course, their value will be eliminated once the robot umpires take over but, in the mean time, it's an area in which players can gain an edge.
One of the ongoing memes between those in analytics and old-school people is the idea that the spreadsheet nerds need to watch the games. Of course, no one has ever suggested teams not watch games; the real value comes in using new data and applying it while watching games. For example, Luhnow said that the Astros incorporate information from their scouts in their "best in class" projection system. He also said that the Astros have five analysts with advanced degrees and they have all spent a lot of time in the clubhouse.
Baseball isn't merely about analytics for teams, though since, with all teams using them, there are fewer advantages available; they also use them to increase fan interest. Squadron noted that predictive analytics can increase fan engagement, including for use in fantasy sports. Taking a page from Adam Silver, I'll also acknowledge that predictive analytics increase the engagement of the gambling audience too.
One of the pieces of new technology that MLB Digital Media showcased, was this tracking technology that would provide data on the routes run by fielders to get to batted balls, designating a straight line as 100% efficient and comparing a fielder's actual route taken. With more information, there will be better ways to evaluate fielders and provide a more comprehensive value to players overall.
There is so much potential with the player tracking data in the sports that are willing to use it.
MORE CONFERENCE NOTES
New England Patriots President Jonathan Kraft: "Statistical analysis in football primarily takes place in the offseason."
Kraft also told a 2000 Draft Day story about how, in the fifth round, Bill Belichick was holding Tom Brady's name card and asking how the Patriots could pass him over, saying he provided too much value, but the Patriots had Drew Bledsoe, didn't need a quarterback, so they used their two fifth-round picks that year on TE Dave Stachelski (one catch for five yards in his career) and DT Jeff Marriott, who never played in the NFL, before relenting and letting Belichick have his guy in the sixth round.
Both Kraft and Calgary Flames President Brian Burke (more on him in the hockey edition) talked about how important it was, in capped leagues, to get useful contributions from players on entry-level contracts.
Former Bulls and Lakers Coach Phil Jackson: "(Rodman) had an engine that never stopped...but it took him to the bars after too."
Jackson also said that the Bulls had to change the guards on Michael Jordan every practice, to build them back up, and cited Warriors PG Steph Curry as a current player that would thrive in his Triangle Offence.
Dean Oliver, ESPN's Director of Production Analytics: "We don't under-estimate the intelligence of the sports fan." Some might argue the World Wide Leader's programming decisions, particularly with respect to this statement, but having a department of Production Analytics shows the kind of effort being made behind the scenes to get accurate and reliable information to the fan.
Oliver was speaking on a Big Data panel with tennis great Mary-Jo Fernandez, who was emphatic in her desire to use data to provide better analysis for tennis viewers. Fernandez showed some interesting plot data on serves by Serena Williams and how when it's compared to others, that Serena's dominance on the serve (and how she achieves it) can be presented in an easy-to-understand visual format. They also use heat maps showing where a player spends most of their time in the match. In the example Fernandez cited, Serena barely left the baseline the whole match.
John Hollinger, who was hired away from ESPN, by the Memphis Grizzlies as their VP of Basketball Operations: "The most important part of basketball analytics is salary cap optimization." This goes back to the point about how much teams ought to be spending on analytics. In a capped league, especially, saving money on the cap affects future moves and what possibilities are open to teams. To think that analytics can't help that process is beyond my comprehension.
Falcons GM Thomas Dimitroff echoed Hollinger's sentiments, saying that you can't separate economics from analytics in sports.
During a golf presentation, there was a an interesting measure of strokes gained and lost, per shot. Presented, in part, by swing coach Sean Foley, the data showed strokes gained and lost for every shot (eg. +0.3 strokes on the tee shot, +0.1 on approach, -0.2 first putt etc.) and it revealed that it's the long game that makes the bigger difference on tour. Foley also used the data to help build up Justin Rose, who was beating himself up over what he thought was a struggling approach game, only to have Foley show him the stats that said Rose was gaining more strokes on his approach than anyone on tour at the time.
On this year's sports betting panel, moderator Jeff Ma expressed his doubts that The Sports Boss, panelist Patrick Donovan, was using a seriously analytical approach when he started talking about getting a feel for a team's emotional level. Any time you're at an analytics conference and someone heads down the road discussing emotion, psychology etc. there are going to be a certain number of eye rolls because no one knows how to quantify that information. My experience has been that kind of thing gets awarded retroactively (ie. winners are determined to have played harder, with more emotion, etc.)
Las Vegas Hotel Superbook Manager Jay Kornegay: "You guys might be very disappointed if I told you how we came up with some of these lines." Kornegay said that they don't use a complicated computer algorithm to determine lines. They have a group of handicappers in-house and they come up with the line as a group but, basically, it's on feel. As Kornegay noted, "We (bookmakers) get paid for being close."
More Kornegay: The Super Bowl line is geared toward public money while the rest of the year is against the sharps.
Check back soon for my piece on Hockey at the 2014 Sloan Sports Analytics Conference!