By now, I'm guessing most basketball fans (and sports fans generally) have seen coverage of the recent Sloan Sports Analytics Conference, and read a few articles about it -- the event has grown significantly since its inception, and Truehoop, Basketball Prospectus and Basketball-Reference, not to mention ESPN and several other blogs have provided excellent coverage of the various panel discussions and research paper presentations this year. This was my first year attending the event, and I was extremely impressed at the quality of the content presented and the general organization of the conference, given how rapidly it has grown in popularity.

I wanted to share some of my thoughts attending the conference, as a first-time attendee passionate about sports and the growing prominence of of analytics. I'll frame some of my observations within the context of having been a part of other conferences in other industries on "hot" topics (some more academically oriented, others more targeted at business management). I have some higher level thoughts about the way the conference was conducted, as well as specific thoughts on the basketball analytics tracks, and some suggestions for what might be good for future events.

The Conference Experience and the Big Panels

I honestly wasn't sure what to expect when I registered for the event back in December, but when I heard that the conference sold out and that it was being moved to the large Boston Convention Center, I originally thought "wow, this is going to be a big event for such a specialized topic". I had originally thought of the event as a more academic and technically oriented conference with some sports management and media people participating. That may have been how things started four years ago, but the growth of interest in analytics -- and its increasingly publicized usage by high profile management at sports teams -- has made the Sloan conference a flashpoint for the discussion of analytics and its intersection with modern sports management.

I spent most of my time with the straight "analytics" talks (basketball analytics, emerging analytics in sports like football and soccer, etc), but there were many interesting "generalist" tracks and panelists covering topics ranging from social media, to alternative storytelling in journalism, to athletes and branding. Panel discussions at other conferences involving high profile speakers aren't always exciting, as it can be hard to get panelists to go in-depth on any topic; meaningful interaction between panelists frequently turns into soundbite type exchanges.

But I was impressed at how engaging many of the panels were, even when the information was high level, such as the "Next Generation Sports Management and Ownership" panel.  The biggest panel of them all and the headliner: "What Geeks Don't Get: The LImits of Moneyball", had a great mix of personalities and Michael Lewis as a genial but intellectually curious and constantly prodding moderator. With Jonathan Kraft and Bill Polian addressing analytics in football, and Mark Cuban and Daryl Morey addressing the basketball side (with Bill Simmons as the media gadfly), there were some very thoughtful exchanges interlaced with humor.

Cuban quoted numbers from a lunchtime presentation on blocked shots to show just how hungry he is for any analytics that can give him a better understanding of the strengths and weaknesses of players on his team, and he and Morey had an interesting exchange on whether "clutch" is something that can be measured and valued (Morey didn't think so, while Cuban felt there was). Even when the information or insights being revealed weren't earth-shattering, it was a pleasure hearing intelligent people think out loud about everything from tactical decision making (the Patriots' oft-argued decision to go for it on 4th down against the Colts) to broader subjects such as how to make analytics accessible to a broader audience.

(This last point about making analytics more accessible -- brought up by Simmons -- is something I'd *really* like to see emphasized at future Sloan events -- maybe as a separate track as the conference will inevitably get bigger).

As far as the actual experience of viewing the panels, the conference center used 3 rooms for the major talks, and a smaller room for the research paper presentations. This has been noted in other recaps, but the organizers made the (reasonable) judgement that certain panels would get the larger rooms. The only problem was that the smallest room (used for the Baseball Analytics and Emerging Analytics talks) was too small, resulting in very crowded standing room only audiences that spilled outside. It did result in the opportunity to eavesdrop on interesting conversations (more on that later), but I assume the organizers will try to manage the spaces for the talks a little better next time.

I'll say this: the crowds in a relatively contained space did make for a certain level of intimacy, even with 1000+ people; it was easy to strike conversations with people. I didn't do too much of it myself, but I appreciated how easy it was to mingle. As the conference gets even bigger, I wonder if this same intimacy can be maintained.

The Basketball Analytics Talks: Best Moments

Some of the basketball discussion, in particular the marquee panel on Basketball Analytics featuring Kevin Pritchard, Dean Oliver, Mike Zarren, Mark Cuban, John Hollinger and Marc Stein, was conducted at a level that both generalists and more statistically minded types could easily digest. The research papers that were presented on subjects like "The Price of Anarchy in Basketball" and regularized plus/minus, were more technical.

Many of the basketball insights from the conference have already been addressed in other articles, but I was particularly drawn to the insights shared in the following talks:

*  A Better Plus/Minus:  Joe Sill presented his paper on "Improved NBA +/- Using Regularization", which I jumped into a little late after sampling the "Emerging Analytics" panel. A detailed explanation of what Sill did to improve on current +/- metrics in the NBA is up on his website, and has already been the subject of some insightful discussion from the excellent Houston Rockets blog Red94. This was the most technical presentation I attended in terms of statistical discussion, with terms like "posterior width" "gaussian prior" and "overfitting" being thrown around -- with ample justification, since Sill was making the case for his more elegant math (as he put it) providing +/- metrics that could provide better predictive value when it comes to analyzing the impact of individuals and team lineups.

The core of Sill's argument was simple enough to follow, and I spotted Mark Cuban and Brent Barry (among others) in the crowd soaking in the discussion. He used examples of players whose impact was either inflated or deflated by current adjusted +/- techniques, such as Brook Lopez (62nd among all players in adjusted +/-), Joe Johnson (309th!!), Chauncey Billups (162nd), and Carmelo Anthony (173rd). Under Sill's regularized +/-, Lopez drops to 173rd, Joe Johnson "improves" to 140th (with Sill providing the qualifier that JJ's defense is what still keeps him down), and Billups (46th) and Melo (37th) rated closer to popular perception of their impact.

Jason Kidd was brought up as someone who still seemed to provide one of those counter intuitive plus/minus rankings that lead mainstream writers like Simmons to dismiss adjusted plus/minus (Kidd is ranked 192nd this season). This led to a fascinating exchange between Sill and Mark Cuban, who defended his team's version of adjusted plus/minus. He pointed out that Kidd's numbers were deflated by mainstream techniques because he was used with a number of the Mavericks' second units to provide stability. Furthermore, the Mavericks weight possessions and scores differently based on their impact on the game (so that if Kidd was part of a lineup that made crucial shots to give the Mavs a win, those points were weighted more than points scored by a lineup early in the game). Also factored into the Mavs' plus/minus measurements were opposing lineups, and how consistent players are.

I enjoyed learning about what Sill is doing, and for my own purposes, I like using plus/minus to evaluate players like Jared Jeffries (72nd in Sill's rankings) who are difficult to measure by conventional metrics because they contribute so little offensively to the box score. There's also a certain story that can be told using plus/minus in evaluating the way lineups are used by teams. One thing that was clear from the conference, though, was that each team has its own "secret sauce" for adjusted plus/minus based on their own valuation of players and lineups, that makes it difficult to provide a truly generalized metric for fans and teams to use as a common language. This was reinforced in the basketball analytics panel by Zarren and Pritchard.

*  The Value of a Blocked Shot: John Huzinga and Sandy Weil presented a superb paper during one of the lunchtime tracks on how not all blocked shots are equal; a good summary of the main points of the presentation is over at ESPN by Peter Keating (with an unfortunate hyperbolic headline about Dwight Howard being "overrated"). Huzinga/Weil looked at blocked shots that come from different situations (after an offensive rebound, live turnover, defensive rebound, live inbounds, dead inbounds), and looked at the points saved based on blocks of layups and jumpers after those events, using play by play data from the last seven years.

The most reported findings from this session were a textbook case for the value of advanced analysis and going beyond the boxscore -- it would be hard for even the most hardened numberphobe to deny the elegance of the case for how Tim Duncan's 149 blocks in 2008 generated more value for his team than Howard's 249 in the same year. At the core of the analysis was the increased value of a blocked shot when it prevented a layup versus a jumpshot, since teams score off layups at a much higher rate -- thus a shot blocker like Jermaine O'Neal (91 percent of blocks are layups) has more value than Brendan Haywood (only 31% of blocks being layups). Haywood's numbers further reinforce a point made by Tom Haberstroh in an article for Hardwood Paroxysm: blocked shots are a poor proxy for basket protection, as the Wizards during Haywood's tenure this season allowed the highest FG% in the league on shots near the rim by opponents. (for Knick fans that made it this far, this means that getting a shot blocker for the sake of blocking shots doesn't necessarily improve the defense)

Other interesting tidbits: only 5 percent of Kendrick Perkins' blocks come on shots that follow an offensive rebound, vs 24 percent of Pau Gasol's blocks -- Huzinga explained that this was due to Gasol's superior quickness in rising for a block (though I'm sure this will hardly quell the "Pau is soft" crowd). Also Rasheed Wallace had the highest number of blocks that ended up out of bounds, while Theo Ratliff had the lowest number. I hope the paper is eventually published online, because I'm sure there is even more insight that can be parsed from the data that may have only glancingly been covered due to time constraints.

The researchers did point out that there were a few factors missing from their research (or perhaps might be subjects for future iterations of the research): 1) the significant intimidation factor generated by a prolific shot blocker like a Dwight Howard, even if he's less efficient than other shot blockers 2) the possible increased fouling that occurs with a shot blocker 3) the possible increase in offensive rebounds given up with shot blocking, such as when teams get weakside blocks but end up in poorer rebounding position.

*  Basketball Analytics - What's Still Missing: The panel of some of the brightest minds in the NBA went down easy, but there was still a certain reticence to share really meaningful information because so much of what teams do presently is proprietary. Cuban was the most forthcoming, and his appeals to the NBA to do more collection of information at a league level to prevent duplication of the same stat-collecting among teams was echoed among the rest of the panel. Information like deflections, locations of blocks, charges, etc would greatly aid teams and fans in getting a more sophisticated understanding of what actually happens in a game, and who's making an impact.

Everyone also agreed that information on injuries is frustratingly scant in the NBA (Morey actually provided an email address to anyone who could provide a breakthrough in that area), and one of the overheard conversations I stumbled onto involved two people working for NBA teams who shared the same frustrations. One of the team representatives said that basketball doesn't require any meaningful reporting of injuries, and that obfuscation is rampant -- the other rep said that they try to work off past injury history of a player  and video analysis when evaluating them, but it was effectively a guessing game. (I should note that I did not engage in any espionage or underhanded eavesdropping -- conversations typically happened right in front of me, as if I didn't exist, given the crowds. I assume the information couldn't have been all that proprietary).

Thoughts For Future Conferences

I chose to spend more time in the foregoing summary talking about more tactical and technical insights that I got from specific talks at the conference, but there was no denying that there was a big buzz at the event arising from the presence of so many bright people seeking and sharing information and ideas. There were a healthy number of sports team executives and personnel at the conference, but there was also media, and plenty of people seeking careers and looking to make connections, and a fair number of bloggers and curious fans.

Knowledge came from the most unexpected places. I chose to focus on the basketball talks and got my money's worth, but I also got serendipitous insights from the likes of Brian Burke and Bill Polian and executives from the English Premier League. There were talks I missed that I wished I could have attended, but thanks to the twitter feeds of bright minds like Jonah Keri (and the super rapid blogging of Truehoop in particular), there was a firehose of information coming from multiple sources.

The appeal of this mix of people and hive of intellectual energy isn't apparent to everyone, based on some of the snarkier reactions that reduce the event to the usual "geek summit" or worse (Exhibit A: "people who haven't won anything, who think they have something to teach us", Exhibit B:  "complexities...that people never see...that...self-promoters without playoff success and championships never talk about at those NBA "Star Trek" conventions in New England" ).

Fortunately, popular mainstream columnists like Simmons and enlightened reporters like Howard Beck of the New York Times understand that it goes beyond paying grudging attention to what the so-called pocket protector crowd has to say. It's about wanting to be smarter, and understanding that there are genuinely new ideas and analysis that produce competitive advantage and can improve our appreciation of individual athletes and teams -- and even sports itself.

There will always be a certain coarse tension between those sympathetic to statistical analysis, and people who believe an intensely quantitative approach understates other intangible factors equally crucial to making good decisions and enjoying sports. But I never felt that tension, or any particular sense of geek exclusivity or triumphalism at the Sloan conference. If anything, the atmosphere seemed particularly welcoming of ideas from people with all sorts of backgrounds, as long as they had an open mind to the possibilities of the new metrics.

Finally, I would *love* to see:

* More from the Recently Converted: I missed the last big talk at Sloan about coaching analytics, though I caught many of the summaries of the talk (here's a good one from Zach Lowe at Celtics Hub), and I think elements of that particular panel should be a part of future Sloan events. Specifically, Avery Johnson and Brent Barry talking about what it was like to incorporate statistical analysis into their way of doing things, and what the pluses and minuses were of doing so; they came at it from the perspective of recent converts, rather than people who had breathing the Kool-Aid all their lives. Insights from recent converts (or on-the-fence skeptics) about what it takes from an organizational standpoint to make advanced statistics culturally acceptable would produce a terrific panel.

I spoke briefly to one team rep whose analytics duties are folded into a broader set of responsibilities touching other departments like marketing and legal; they were thinly staffed, to put it generously. Teams like Houston have a rumored reputation for having substantial personnel devoted to analytics, but some teams scuffle with much less. It may be too hard to pry this information from teams, but discussing how an infrastructure for analytics can be assembled within teams (in multiple sports) would be interesting.

* More From Players and Even Skeptics: Maybe future Sloan conferences can push the envelope (at the risk of generating some tension) by inviting scouts, or people who've used statistics but keep them at arm's length and have war stories to share about why they prefer to do things the way they do. Maybe some players open to digesting the most sophisticated scouting reports, like Shane Battier (if Morey would allow it), would be insightful. The important thing, of course, would be to try and invite people with some level of open-mindedness, so that it doesn't become an uncomfortable, unproductive standoff between "oldheads" and stats dudes.

* More From Media and Blogs:  What are blogs and mainstream media doing to make advanced analytics part of the more general sports discussion, without alienating fans? There was a great article on Baseball Prospectus by Jon Sciambi about this topic recently (the comments are just as worthwhile as the article itself), and I would love to see this topic explored further. Could Doris Burke or Jeff Van Gundy share some numbers about, say, Brandon Roy's passing and shooting tendencies with him using some advanced metrics and expect to get a thoughtful exchange?

*  More From Sports Leagues:  A little more on what leagues are doing to make more advanced information available (to the degree they're interested in doing so) would be terrific. John Shuhmann at NBA.com has a regular column with advanced analyses on a variety of topics, like clutchness; having John or other representatives of the NBA talk about the tools they have, and how they may be trying to collect more information for teams beyond the standard box score, would certainly interest me.

There are many directions the Sloan conference can go, and I'm looking forward to attending future installments. I would recommend that anyone who cares about sports should try and do the same.